Prior to this PR, the HOMEDRIVE environment variable was used to
detect what drive we are operating in. This variable is not available
for service account logins (like what is used for CI), so switch to
extracting the drive from PROGRAMFILES (which is more-widely defined).
On Windows, several commonly available system tools for decompression
are unreliable (gz/bz2/xz). This commit refactors `decompressor_for`
to call out to a Windows or Unix-specific method:
* The decompressor_for_nix method behaves the same as before and
generally treats the Python/system support options for decompression
as interchangeable (although avoids using Python's built-in tar
support since that has had issues with permissions).
* The decompressor_for_win method can only use Python support for
gz/bz2/xz, although for a tar.gz it does use system support for
untar (after the decompression step). .zip uses the system tar
utility, and .Z depends on external support (i.e. that the user
has installed 7zip).
A naming scheme has been introduced for the various _decompression
methods:
* _system_gunzip means to use a system tool (and fail if it's not
available)
* _py_gunzip means to use Python's built-in support for decompressing
.gzip files (and fail if it's not available)
* _gunzip is a method that can do either
This is a refactor of Spack's stand-alone test process to be more spack- and pytest-like.
It is more spack-like in that test parts are no longer "hidden" in a package's run_test()
method and pytest-like in that any package method whose name starts test_
(i.e., a "test" method) is a test part. We also support the ability to embed test parts in a
test method when that makes sense.
Test methods are now implicit test parts. The docstring is the purpose for the test part.
The name of the method is the name of the test part. The working directory is the active
spec's test stage directory. You can embed test parts using the test_part context manager.
Functionality added by this commit:
* Adds support for multiple test_* stand-alone package test methods, each of which is
an implicit test_part for execution and reporting purposes;
* Deprecates package use of run_test();
* Exposes some functionality from run_test() as optional helper methods;
* Adds a SkipTest exception that can be used to flag stand-alone tests as being skipped;
* Updates the packaging guide section on stand-alone tests to provide more examples;
* Restores the ability to run tests "inherited" from provided virtual packages;
* Prints the test log path (like we currently do for build log paths);
* Times and reports the post-install process (since it can include post-install tests);
* Corrects context-related error message to distinguish test recipes from build recipes.
fixes#22341
Using double quotes creates issues with shell variable substitutions,
in particular when the manifest has "definitions:" in it. Use single
quotes instead.
Add a "require" directive to packages, which functions exactly like
requirements specified in packages.yaml (uses the same fact-generation
logic); update both to allow making the requirement conditional.
* Packages may now use "require" to add constraints. This can be useful
for something like "require(%gcc)" (where before we had to add a
conflict for every compiler except gcc).
* Requirements (in packages.yaml or in a "require" directive) can be
conditional on a spec, e.g. "require(%gcc, when=@1.0.0)" (version
1.0.0 can only build with gcc).
* Requirements may include a message which clarifies why they are needed.
The concretizer assigns a high priority to errors which generate these
messages (in particular over errors for unsatisfied requirements that
do not produce messages, but also over a number of more-generic
errors).
## Version types, parsing and printing
- The version classes have changed: `VersionBase` is removed, there is now a
`ConcreteVersion` base class. `StandardVersion` and `GitVersion` both inherit
from this.
- The public api (`Version`, `VersionRange`, `ver`) has changed a bit:
1. `Version` produces either `StandardVersion` or `GitVersion` instances.
2. `VersionRange` produces a `ClosedOpenRange`, but this shouldn't affect the user.
3. `ver` produces any of `VersionList`, `ClosedOpenRange`, `StandardVersion`
or `GitVersion`.
- No unexpected type promotion, so that the following is no longer an identity:
`Version(x) != VersionRange(x, x)`.
- `VersionList.concrete` now returns a version if it contains only a single element
subtyping `ConcreteVersion` (i.e. `StandardVersion(...)` or `GitVersion(...)`)
- In version lists, the parser turns `@x` into `VersionRange(x, x)` instead
of `Version(x)`.
- The above also means that `ver("x")` produces a range, whereas
`ver("=x")` produces a `StandardVersion`. The `=` is part of _VersionList_
syntax.
- `VersionList.__str__` now outputs `=x.y.z` for specific version entries,
and `x.y.z` as a short-hand for ranges `x.y.z:x.y.z`.
- `Spec.format` no longer aliases `{version}` to `{versions}`, but pulls the
concrete version out of the list and prints that -- except when the list is
is not concrete, then is falls back to `{versions}` to avoid a pedantic error.
For projections of concrete specs, `{version}` should be used to render
`1.2.3` instead of `=1.2.3` (which you would get with `{versions}`).
The default `Spec` format string used in `Spec.__str__` now uses
`{versions}` so that `str(Spec(string)) == string` holds.
## Changes to `GitVersion`
- `GitVersion` is a small wrapper around `StandardVersion` which enriches it
with a git ref. It no longer inherits from it.
- `GitVersion` _always_ needs to be able to look up an associated Spack version
if it was not assigned (yet). It throws a `VersionLookupError` whenever `ref_version`
is accessed but it has no means to look up the ref; in the past Spack would
not error and use the commit sha as a literal version, which was incorrect.
- `GitVersion` is never equal to `StandardVersion`, nor is satisfied by it. This
is such that we don't lose transitivity. This fixes the following bug on `develop`
where `git_version_a == standard_version == git_version_b` does not imply
`git_version_a == git_version_b`. It also ensures equality always implies equal
hash, which is also currently broken on develop; inclusion tests of a set of
versions + git versions would behave differently from inclusion tests of a
list of the same objects.
- The above means `ver("ref=1.2.3) != ver("=1.2.3")` could break packages that branch
on specific versions, but that was brittle already, since the same happens with
externals: `pkg@1.2.3-external` suffixes wouldn't be exactly equal either. Instead,
those checks should be `x.satisfies("@1.2.3")` which works both for git versions and
custom version suffixes.
- `GitVersion` from commit will now print as `<hash>=<version>` once the
git ref is resolved to a spack version. This is for reliability -- version is frozen
when added to the database and queried later. It also improves performance
since there is no need to clone all repos of all git versions after `spack clean -m`
is run and something queries the database, triggering version comparison, such
as potentially reuse concretization.
- The "empty VerstionStrComponent trick" for `GitVerison` is dropped since it wasn't
representable as a version string (by design). Instead, it's replaced by `git`,
so you get `1.2.3.git.4` (which reads 4 commits after a tag 1.2.3). This means
that there's an edge case for version schemes `1.1.1`, `1.1.1a`, since the
generated git version `1.1.1.git.1` (1 commit after `1.1.1`) compares larger
than `1.1.1a`, since `a < git` are compared as strings. This is currently a
wont-fix edge case, but if really required, could be fixed by special casing
the `git` string.
- Saved, concrete specs (database, lock file, ...) that only had a git sha as their
version, but have no means to look the effective Spack version anymore, will
now see their version mapped to `hash=develop`. Previously these specs
would always have their sha literally interpreted as a version string (even when
it _could_ be looked up). This only applies to databases, lock files and spec.json
files created before Spack 0.20; after this PR, we always have a Spack version
associated to the relevant GitVersion).
- Fixes a bug where previously `to_dict` / `from_dict` (de)serialization would not
reattach the repo to the GitVersion, causing the git hash to be used as a literal
(bogus) version instead of the resolved version. This was in particularly breaking
version comparison in the build process on macOS/Windows.
## Installing or matching specific versions
- In the past, `spack install pkg@3.2` would install `pkg@=3.2` if it was a
known specific version defined in the package, even when newer patch releases
`3.2.1`, `3.2.2`, `...` were available. This behavior was only there because
there was no syntax to distinguish between `3.2` and `3.2.1`. Since there is
syntax for this now through `pkg@=3.2`, the old exact matching behavior is
removed. This means that `spack install pkg@3.2` constrains the `pkg` version
to the range `3.2`, and `spack install pkg@=3.2` constrains it to the specific
version `3.2`.
- Also in directives such as `depends_on("pkg@2.3")` and their when
conditions `conflicts("...", when="@2.3")` ranges are ranges, and specific
version matches require `@=2.3.`.
- No matching version: in the case `pkg@3.2` matches nothing, concretization
errors. However, if you run `spack install pkg@=3.2` and this version
doesn't exist, Spack will define it; this allows you to install non-registered
versions.
- For consistency, you can now do `%gcc@10` and let it match a configured
`10.x.y` compiler. It errors when there is no matching compiler.
In the past it was interpreted like a specific `gcc@=10` version, which
would get bootstrapped.
- When compiler _bootstrapping_ is enabled, `%gcc@=10.2.0` can be used to
bootstrap a specific compiler version.
## Other changes
- Externals, compilers, and develop spec definitions are backwards compatible.
They are typically defined as `pkg@3.2.1` even though they should be
saying `pkg@=3.2.1`. Spack now transforms `pkg@3` into `pkg@=3` in those cases.
- Finally, fix strictness of `version(...)` directive/declaration. It just does a simple
type check, and now requires strings/integers. Floats are not allowed because
they are ambiguous `str(3.10) == "3.1"`.
`spack buildcache create` is a misnomer cause it's the only way to push to
an existing buildcache (and it in fact calls binary_distribution.push).
Also we have `spack buildcache update-index` but for create the flag is
`--rebuild-index`, which is confusing (and also... why "rebuild"
something if the command is "create" in the first place, that implies it
wasn't there to begin with).
So, after this PR, you can use either
```
spack buildcache create --rebuild-index
```
or
```
spack buildcache push --update-index
```
Also, alias `spack buildcache rebuild-index` to `spack buildcache
update-index`.
Spack never parsed `nagfor` linker arguments put on the compiler line:
```
nagfor -Wl,-Wl,,-rpath,,/path
````
so, let's continue not attempting to parse that.
`buildcache create --rel`: deprecate this because there is no point in
making things relative before tarballing; on install you need to expand
`$ORIGIN` / `@loader_path` / relative symlinks anyways because some
dependencies may actually be in an upstream, or have different
projections.
`buildcache install --allow-root`: this flag was propagated through a
lot of functions but was ultimately unused.
This switches the default Make build type to `build_type=Release`.
This offers:
- higher optimization level, including loop vectorization on older GCC
- adds NDEBUG define, which disables assertions, which could cause speedups if assertions are in loops etc
- no `-g` means smaller install size
Downsides are:
- worse backtraces (though this does NOT strip symbols)
- perf reports may be useless
- no function arguments / local variables in debugger (could be of course)
- no file path / line numbers in debugger
The downsides can be mitigated by overriding to `build_type=RelWithDebInfo` in `packages.yaml`,
if needed. The upside is that builds will be MUCH smaller (and faster) with this change.
---------
Co-authored-by: Gregory Becker <becker33@llnl.gov>
* Vendor ruamel.yaml v0.17.21
* Add unit test for whitespace regression
* Add an abstraction layer in Spack to wrap ruamel.yaml
All YAML operations are routed through spack.util.spack_yaml
The custom classes have been adapted to the new ruamel.yaml
class hierarchy.
Fixed line annotation issue in "spack config blame"
This ensures that:
a) no externals are added to the tarball metadata file
b) no externals are added to the prefix to prefix map on install, also
for old tarballs that did include externals
c) ensure that the prefix -> prefix map is always string to string, and
doesn't contain None in case for some reason a hash is missing
* Disable module generation by default (#35564)
a) It's used by site administrators, so it's niche
b) If it's used by site administrators, they likely need to modify the config anyhow, so the default config only serves as an example to get started
c) it's too arbitrary to enable tcl, but disable lmod
* Remove leftover from old module file schema
* Warn if module file config is detected and generation is disabled
---------
Co-authored-by: Harmen Stoppels <harmenstoppels@gmail.com>
Change the signature of the Environment.__init__ method to have
a single argument, i.e. the directory where the environment manifest
is located. Initializing that directory is now delegated to a function
taking care of all the error handling upfront. Environment objects
require a "spack.yaml" to be available to be constructed.
Add a class to manage the environment manifest file. The environment
now delegates to an attribute of that class the responsibility of keeping
track of changes modifying the manifest. This allows simplifying the
updates of the manifest file, and helps keeping in sync the spec lists in
memory with the spack.yaml on disk.
Spack comes to a crawl post-install of nvhpc, which is partly thanks to
this post install hook which has a lot of redundancy, and isn't correct.
1. There's no need to store "type" because that _is_ "mode".
2. There are more file types than "symlink", "dir", "file".
3. Don't checksum device type things
4. Don't run 3 stat calls (exists, stat, isdir/islink), but one lstat
call
5. Don't read entire files into memory
I also don't know why `spack.crypto` wasn't used for checksumming, but I
guess it's too late for that now. Finally md5 would've been the faster
algorithm, which would've been fine given that a non cryptographicall
checksum was used anyways.
If we modify both Path and PATH, on Windows they will clobber one
another. This PR updates the shell modification logic to automatically
convert variable names to upper-case on Windows.
- [x] Replace `version(ver, checksum=None, **kwargs)` signature with
`version(ver, checksum=None, *, sha256=..., ...)` explicitly listing all arguments.
- [x] Fix various issues in packages:
- `tags` instead of `tag`
- `default` instead of `preferred`
- `sha26` instead of `sha256`
- etc
Also, use `sha256=...` consistently.
Note: setting `sha256` currently doesn't validate the checksum length, so you could do
`sha256="a"*32` and it would get checked as `md5`... but that's something for another PR.
This means that `spack install` will now build the minimal set of packages
required to install the root(s).
To opt out of build edge pruning, use `spack install --include-build-deps`.
Co-authored-by: Harmen Stoppels <harmenstoppels@gmail.com>
Other tools like git support `GIT_EDITOR` which takes higher precedence than the
standard `VISUAL` or `EDITOR` variables. This adds similar support for Spack, in the
`SPACK_EDITOR` env var.
- [x] consolidate editor code from hooks into `spack.util.editor`
- [x] add more editor tests
- [x] add support for `SPACK_EDITOR`
- [x] add a documentation section for controlling the editor and reference it
Code from `spack.util.editor` was duplicated into our licensing hook in #11968. We
really only want one place where editor search logic is implemented. This consolidates
the logic into `spack.util.editor`, including a special case to run `gvim` with `-f`.
- [x] consolidate editor search logic in spack.util.editor
- [x] add tests for licensing case, where `Executable` is used instead of `os.execv`
- [x] make `_exec_func` argument of `editor()` into public `exec_fn` arg
- [x] add type annotations
fixes#36628
Fix using compilers that declare "target: any" in their
configuration. This should happen only on Cray with the
module based programming environment.
* Simplify test/cmd/ci.py::test_ci_generate_with_custom_scripts
* Rearrange the build-job logic in generate_gitlab_ci_yaml
* Preserve all unknown attributes in build jobs
* Slip tests for custom attributes in the tests for other job types
* Support custom artifacts
* [@spackbot] updating style on behalf of blue42u
* Don't bother sorting needs
---------
Co-authored-by: blue42u <blue42u@users.noreply.github.com>
Other tools like git support `GIT_EDITOR` which takes higher precedence than the
standard `VISUAL` or `EDITOR` variables. This adds similar support for Spack, in the
`SPACK_EDITOR` env var.
- [x] consolidate editor code from hooks into `spack.util.editor`
- [x] add more editor tests
- [x] add support for `SPACK_EDITOR`
- [x] add a documentation section for controlling the editor and reference it
Code from `spack.util.editor` was duplicated into our licensing hook in #11968. We
really only want one place where editor search logic is implemented. This consolidates
the logic into `spack.util.editor`, including a special case to run `gvim` with `-f`.
- [x] consolidate editor search logic in spack.util.editor
- [x] add tests for licensing case, where `Executable` is used instead of `os.execv`
- [x] make `_exec_func` argument of `editor()` into public `exec_fn` arg
- [x] add type annotations
* lmod modules: allow users to remove items from hierarchy per-spec
This allows MPI wrappers that depend on MPI to be removed from the MPI portion of
the hierarchy and be made available when the appropriate compiler is loaded.
module load gcc
module load mpi-wrapper # implicitly loads mpi
module load hdf5
This allows users to treat an mpi wrapper like an mpi program
This commit changes the environment modifications class to escape
strings with double quotes instead of single quotes.
Single quotes prevent the expansion of enviornment variables that are
nested within environment variable definitions.
Fixes#36689
- The "base" builder class should be last in the MRO
- `filter_compiler_wrappers` needs to be moved to builders
- Decorating a function from a mixin class require using
the correct metaclass for the mixin
This aims to resolve#34164 by resolving the <include-fragment> tags
that GitHub has started using for their release pages, see
https://github.github.io/include-fragment-element/.
This feels a bit hacky but intended as a starting point for discussion.
After reading a page during spidering, it first parses for
include-fragments, gets them all, and treats them all as separate pages.
Then it looks for href links in both the page itself and the fragments.
Co-authored-by: Alec Scott <alec@bcs.sh>
MSVC compilers rely on vcvars environment setup scripts to establish
build environement variables neccesary for all projects to build
successfully. Prior to this we were only piping LIB, INCLUDE, and PATH
change through.
Instead we need to propegate all changes to the env variables made by
VCVARs in order to establish robust support for the MSVC compiler.
This most significantly impacts projects that need to be build with
NMake and MSBuild
* CI: Fixup docs for bootstrap.
* CI: Add compatibility shim
* Add an update method for CI
Update requires manually renaming section to `ci`. After
this patch, updating and using the deprecated `gitlab-ci` section
should be possible.
* Fix typos in generate warnings
* Fixup CI schema validation
* Add unit tests for legacy CI
* Add deprecated CI stack for continuous testing
* Allow updating gitlab-ci section directly with env update
* Make warning give good advice for updating gitlab-ci
* Fix typo in CI name
* Remove white space
* Remove unneeded component of deprected-ci
* Extract a method to warn when the manifest is not up-to-date
* Extract methods to update the repository and ensure dir exists
* Simplify further the write method, add failing unit-test
* Fix the function computing YAML equivalence between two instances
Previously `spack -e bla config update <section>` would treat the
environment config scope as standard config file instead of a single
file config scope. This fixes that.
When app is uninstalled, if it matches a default, then remove the
default symlink targeting its modulefile.
Until now, when a default were uninstalled, the default symlink were
left pointing to a nonexistent modulefile.
- Update default image to Ubuntu 22.04 (previously was still Ubuntu 18.04)
- Optionally use depfiles to install the environment within the container
- Allow extending Dockerfile Jinja2 template
- Allow extending Singularity definition file Jinja2 template
- Deprecate previous options to add extra instructions
* Reduce effort on grounding by employing cardinality constraints
If we use a cardinality constraint instead of a rule
using pair of values, we'll end up grounding 1 rule
instead of all the possible pair combinations of the
allowed values.
* Display all errors from concretization, instead of just one
If clingo produces multiple "error" facts, we now print all
of them in the error message. Before we were printing just
the one with the least priority.
Consolidate a few common patterns in concretize.lp to ensure
that certain node attributes have one and only one value
assigned.
All errors are displayed, so use a single criterion
instead of three.
* Account for weights in concretize.lp
To recover the optimization order we had before, account
for weights of errors when minimizing.
The priority is mapped to powers of 10, so to effectively
get back the same results as with priorities.
When generating modulefile, correctly detect software installation asked
by user as explicit installation.
Explicit installation status were previously fetched from database
record of spec, which was only set after modulefile generation.
Code is updated to pass down the explicit status of software
installation to the object that generates modulefiles.
Fixes#34730.
Fixes#12105.
A value for the explicit argument has to be set when creating a new
installation, but for operations on existing installation, this value is
retrieved from database. Such operations are: module rm, module refresh,
module setdefaults or when get_module function is used.
Update on the way tests that mimics an installation, thus explicit
argument has to be set under such situation.
Add `config:stage_name` which is a Spec format string that can
customize the names of stages created by Spack. This was primarily
created to allow generating shorter stage names on Windows (along
with `config:build_stage`, this can be used to create stages with
short absolute paths).
By default, this is not set and the prior name stage format is used.
This also removes the username component that is always added to
Stage paths on Windows (if users want to include this, they can
add it to the `build_stage`).
* compiler wrapper: fix -Xlinker parsing
* handle the case of -rpath without value; avoid that we drop the flag
* also handle the -Xlinker -rpath -Xlinker without further args case...
* fix test
* get rid of global $rp var, reduce branching
- [x] Specs that define 'new' versions in the require: section need to generate associated facts to indicate that those versions are valid.
- [x] add test to verify success with unknown versions.
- [x] remove unneeded check that was leading to extra complexity and test
failures (at this point, all `hash=version` does not require listing out that version
in `packages.yaml`)
- [x] unique index for origin (dont reuse 0)
Co-authored-by: Peter Josef Scheibel <scheibel1@llnl.gov>
Add a `find_first` method that locates one instance of a file
that matches a specified pattern by recursively searching a directory
tree. Unlike other `find` methods, this only locates one file at most,
so can use optimizations that avoid searching the entire tree:
Typically the relevant files are at low depth, so it makes sense to
locate files through iterative deepening and early exit.
Update tcl and lmod modulefile template to provide more information on
help message (name, version and target) like done on whatis for lmod
modulefiles.
Adapt tcl and lmod modulefile templates to generate append-path or
remove-path commands in modulefile when respectively append_flags or
remove_flags commands are defined in package for run environment.
Fixes#10299.
Simplify environment modification block in modulefile Tcl template by
always setting a path delimiter to the prepend-path, append-path and
remove-path commands.
Remove --delim option to the setenv command as this command does not
allow such option.
Update test_prepend_path_separator test to explicitly check the 6
path-like commands that should be present in generated modulefile.
fixes#36339
We were missing a rule that enforced a match between
the `node_compiler` and the compiler used to satisfy
a requirement.
Fix compiler with custom, made up version too
* Specs that define 'new' versions in the require: section need to generate
associated facts to indicate that those versions are valid.
* add test to verify success with unknown versions.
Since environment-modules has support for autoloading since 4.2,
and Spack-builds of it enable it by default, use the same autoload
default for tcl as lmod.
If you have a "require:" section in your packages config, and you
use it to specify a list of requirements, the list elements can
now include strings (before this, each element in the list had to
be a `one_of` or `any_of` specification, which is awkward if you
wanted to apply just one spec with no alternatives).
Example one:
```
spack install --add x y z
```
is equivalent to
```
spack add x y z
spack concretize
spack install --only-concrete
```
where `--only-concrete` installs without modifying spack.yaml/spack.lock
Example two:
```
spack install
```
concretizes current spack.yaml if outdated and installs all specs.
Example three:
```
spack install x y z
```
concretizes current spack.yaml if outdated and installs *only* concrete
specs in the environment that match abstract specs `x`, `y`, or `z`.
The `ignore` parameter was only used for `spack activate/deactivate`, and it isn't used
by Spack Environments which have their own handling of file conflicts. We should remove it.
Everything that handles `ignore=` was removed in #29317 and included in 0.19, when we
removed `spack activate` and `spack deactivate` in favor of environments. So all of these
usages removed here were already being ignored by Spack.
Adapt tcl modulefile template to call "module load" on autoload
dependency without testing if this dependency is already loaded or not.
The is-loaded test is not necessary, as module commands know how to cope
with an already loaded module. With environment-modules 4.2+ (released
in 2018) it is also important to have this "module load" command even if
dependency is already loaded in order to record that the modulefile
declares such dependency. This is important if you want to keep a
consistent environment when a dependent module is unloaded.
The "Autoloading" verbose message is also removed as recent module
commands will report such information to the user (depending on the
verbosity configured for the module command).
Such change has been test successfully with Modules 3.2 (EL7), 4.5 (EL8)
and 5.2 (latest) and also with Lmod 7 and 8 (as it is mentionned in
Spack docs that Lmod can be used along with tcl modules). Dependencies
are correctly loaded or unloaded, whether they are loaded/unloaded or
not.
This change fixes Tcl quoting issue introduced in #32853.
Fixes#19155.
In the Windows filesystem logic for creating a symlink, we intend to
fall back to a copy when the symlink cannot be created (for some
configuration settings on Windows it is not possible for the user
to create a symlink). It turns out we were overly-broad in which
exceptions lead to this fallback, and the subsequent copy would
also fail: at least one case where this occurred is when we
attempted to create a symlink that already existed.
The updated logic expressly avoids falling back to a copy when the
file/symlink already exists.
* ASP-based solver: use satisfies instead of intersects
They are semantically equivalent for concrete versions,
but the GitVersion.intersects implementation is buggy
* Mitigation for git version bug
fixes#36134
This commit works around the issue in #36134, by using
GitVersion.satisfies instead of GitVersion.intersects
There are still underlying issues when trying to infer the
"reference version" when no explicit one is given, but:
1. They are not reproducible with our synthetic repo
2. They occur only when the `git.<xxx>` form of Git version
is used
Here we just work around the user facing issue and ensure
the tests are correct with our synthetic repository.
This PR does 2 unrelated things:
1. It changes the encoding of the compilers
2. It tweaks the heuristic for the solves in a0d88179074f81d13a3fad629a43d86334e7f982
Both were initially motivated by trying to get a performance gain but, while 2 showed significant speed-ups[^1], 1 instead didn't. I kept it anyhow, since I think the code related to compilers is more consolidated with the new encoding and we might get some performance improvement out of it if we can base our errors on the `node_compiler(Package, CompilerID)` atoms instead of `attrs`.
[^1]: In general the changes in the heuristic brought a ~10% speed-up on the tests I did. I'll post detailed results below.
Add a warning about compilers.yaml that is triggered if there are multiple compilers with the same spec, os and
target (since they can't be selected by users with the spec syntax only).
Since GPG clear-sign cannot deal with lines longer than 19995 characters
and doesn't even error but simply truncates those linese (don't ask me
why...), we have to be careful not to hit that line limit when reducing
the filesize.
So, instead this PR sets the indent level to 0 and drops the whitespace
after `: `, which still reduces file size by 50% or so.
Adds builders appropriate for building these packages on Windows.
It is intended that builds on other platforms are unaffected (e.g.
they build with Autotools as before on Linux).
... and use colors in disambiguate message for clarity.
This commit avoids the loop:
```
for root in roots:
for dep in deps(root):
...
```
instead it ensures each node is visited once and only once.
Also adds a small optimization when searching for concrete specs, since
we can assume uniqueness of dag hash, so it's fine to early exit.
This adds a new mode for `concretizer:reuse` called `dependencies`,
which only reuses dependencies. Currently, `spack install foo` will
reuse older versions of `foo`, which might be surprising to users.
* CI configuration boilerplate reduction and refactor
Configuration:
- New notation for list concatenation (prepend/append)
- New notation for string concatenation (prepend/append)
- Break out configuration files for: ci.yaml, cdash.yaml, view.yaml
- Spack CI section refactored to improve self-consistency and
composability
- Scripts are now lists of lists and/or lists of strings
- Job attributes are now listed under precedence ordered list that are
composed/merged using Spack config merge rules.
- "service-jobs" are identified explicitly rather than as a batch
CI:
- Consolidate common, platform, and architecture configurations for all CI stacks into composable configuration files
- Make padding consistent across all stacks (256)
- Merge all package -> runner mappings to be consistent across all
stacks
Unit Test:
- Refactor CI module unit-tests for refactor configuration
Docs:
- Add docs for new notations in configuration.rst
- Rewrite docs on CI pipelines to be consistent with refactored CI
workflow
* Script verbose environ, dev bootstrap
* Port #35409
When untouched spec pruning is enabled, specs possibly affected
by a change cannot be pruned from a pipeline.
Previously spack looked at all specs matching changed package
names, and traversed dependents of each, all the way to the
environment root, to compute the set of environment specs
possibly affected by a change (and thus, not candidates for
pruning).
With this PR, when untouched spec pruning is enabled, a new
environment variable can control how far towards the root spack
traverses to compute the set of specs possibly affected by a
change. SPACK_UNTOUCHED_PRUNING_DEPENDENT_DEPTH can be set
to any numeric value before the "spack ci generate" command
is called to control this traversal depth parameter. Setting
it to "0" traverses only touched specs, setting it to "1"
traverses only touched specs and their direct dependents, and
so on. Omitting the variable results in the previous behavior
of traversing all the way to the root. Setting it to a negative
value means no traversal is done, and always yields an empty
set of possibly affected specs (which would result in the max
pruning possible).
Currently `spack buildcache create` creates compressed tarballs that
differ between each invocation, thanks to:
1. The gzip header containing mtime set to time.time()
2. The generated buildinfo file which has a different mtime every time.
To avoid this, you have to explicitly construct GZipFile yourself, since
the Python API doesn't expose the mtime arg, and we have to manually
create the tarinfo object for the buildinfo metadata file.
Normalize mode: regular files & hardlinks executable by user, dirs, symlinks: set 0o755 permissions in tarfile; other files use 0o644
This commit formalizes `satisfies(lhs, rhs, strict=True/False)`
and splits it into two functions: `satisfies(lhs, rhs)` and
`intersects(lhs, rhs)`.
- `satisfies(lhs, rhs)` means: all concrete specs matching the
left hand side also match the right hand side
- `intersects(lhs, rhs)` means: there exist concrete specs
matching both lhs and rhs.
`intersects` now has the property that it's commutative,
which previously was not guaranteed.
For abstract specs, `intersects(lhs, rhs)` implies that
`constrain(lhs, rhs)` works.
What's *not* done in this PR is ensuring that
`intersects(concrete, abstract)` returns false when the
abstract spec has additional properties not present in the
concrete spec, but `constrain(concrete, abstract)` will
raise an error.
To accomplish this, some semantics have changed, as well
as bugfixes to ArchSpec:
- GitVersion is now interpreted as a more constrained
version
- Compiler flags are interpreted as strings since their
order is important
- Abstract specs respect variant type (bool / multivalued)
Two fixes:
1. `-Wl,a,b,c,d` is a comma separated list of linker arguments, we
incorrectly assume key/value pairs, which runs into issues with for
example `-Wl,--enable-new-dtags,-rpath,/x`
2. `-Xlinker,xxx` is not a think, so it shouldn't be parsed.
Currently, if two compilers with the same spec differ on the flags, the concretizer will:
1. mix both sets of flags for the spec in the ASP program
2. error noting that the set of flags from the compiler (both of them) doesn't match the set from the lower priority compiler
This PR fixes both -- only flags from the highest priority compiler with a given spec are considered.
`mypy` only understands `sys.platform == "win32"`, not indirect assignments of that
value to things like `is_windows`. If we don't use the accepted platform checks, `mypy`
registers many Windows-only symbols as not present on Linux, when it should skip the
checks for platform-specific code.
Update `spack.util.environment` to remove legacy idioms.
* Remove kwargs from method signature and use a class for traces
* Uppercase a few global variables
* spack.util.environment: add type-hints
* Improve docstrings
* Fixed most style issues reported by pylint
a shared library /lib64/libcxi.so, which seems to also appear on other
non-slingshot systems. This patch also checks to make sure that there
is a Cray programming enviornment in /opt/cray/pe in addition to the
shared library.
if dump file existed it was not truncating the file, resulting in
a file with unaltered filesize, with the new content at the beginning,
"padded" with the tail of the old content, since the new content was
not enough to overwrite it.
`colify` is an old module in Spack that still uses `**kwargs` liberally.
We should be more explicit. Doing this eliminates the need for many
checks (can't pass the wrong arg if it isn't allowed) and makes the
function documentation more clear.
Fixes a bug introduced in 44ed0de8c0
where the push method of binary_distribution now takes named args
include_root and include_depedencies, to avoid the **kwarg hole.
But the call site wasn't update and we passed a dict of keys/values instead
of arguments, which resulted in a call like this:
```
push(include_root={"include_root": True, "include_dependencies": False})
```
This commit fixes that, and adds a test to see if we push the correct packages.
This error shows up a lot, typically it's harmless because an error
happened before the source build even started, in which case we don't
have build logs to copy. So, warn instead of error, cause it distracts
from the actual CI error.
Currently we attempt to setup the build environment even when
dependencies are not installed, which typically results in error while
searching for libraries or executables in a dependency's prefix.
With this change, we get a more user friendly error:
```
$ spack build-env perl
==> Error: Not all dependencies of perl are installed, cannot setup build environment:
- qpj6dw5 perl@5.36.0%apple-clang@14.0.0+cpanm+open+shared+threads build_system=generic arch=darwin-ventura-m1
- jq2plbe ^berkeley-db@18.1.40%apple-clang@14.0.0+cxx~docs+stl build_system=autotools patches=26090f4,b231fcc arch=darwin-ventura-m1
...
$ echo $?
1
```
* Allow users to specify root env dir
Environments managed by spack have some advantages over anonymous Environments
but they are tucked away inside spack's directory tree. This PR gives
users the ability to specify where the environments should live.
See #32823
This is also taken as an opportunity to ensure that all references are to "managed environments",
rather than "named environments". Prior to this PR some references to the latter persisted.
Co-authored-by: Tom Scogland <scogland1@llnl.gov>
Co-authored-by: Tamara Dahlgren <35777542+tldahlgren@users.noreply.github.com>
Co-authored-by: Gregory Becker <becker33@llnl.gov>
The call:
```
x.satisfies(y[, strict=False])
```
is commutative, and tests non-empty intersection, whereas:
```
x.satsifies(y, strict=True)
```
is not commutative, and tests set-inclusion.
There are 2 fast paths. When strict=False both self and other need to
be concrete, when strict=True we can optimize when other is concrete.
Spack generally ignores file-file projection clashes in environment
views, but would eventually error when linking the `.spack` directory
for two specs of the same package.
This leads to obscure errors where users have no clue what the issue is
and how to fix it. On top of that, the error comes very late, since it
happens when the .spack dir contents are linked (which happens after
everything else)
This PR improves that by doing a quick check ahead of time if clashes
are going to be anticipated (by simply checking for clashes in the
projection of each spec's .spack metadir). If there are clashes, a
human-readable error is thrown which shows two of the conflicting specs,
and tells users to user unify:true, view:false, or set up custom
projections.
The checksum exception was not detailed enough and not reraised when using cache only, resulting in useless error messages.
Now it dumps the file path, expected
hash, computed hash, and the downloaded file summary.
* Style: black 23, skip magic trailing commas
* isort should use same line length as black
* Fix unused import
* Update version of black used in CI
* Update new packages
* Update new packages
Specs that did not contribute any files to an env view caused a problem
where zip(specs, files grouped by prefix) got "out of sync", causing the
wrong merge map to be passed to a package's `add_files_to_view`, which
specifically caused an issue where *sometimes* bin/python ended up as a
symlink instead of a copy.
One such example is kokkos + kokkos-nvcc-wrapper, as the latter package
only provides the file bin/nvcc_wrapper, which is also added to view by
kokkos, causing kokkos-nvcc-wrapper to contribute 0 files.
The test feels a bit contrived, but it captures the problem... pkg a is
added first and has 0 files to contribute, pkg b adds a single file, and
we check if pkg b receives a merge map (and a does not).
`spack gc` removes build deps of explicitly installed specs, but somehow
if you take one of the specs that `spack gc` would remove, and feed it
to `spack uninstall /<hash>` by hash, it complains about all the
dependents that still rely on it.
This resolves the inconsistency by only following run/link type deps in
spack uninstall.
That way you can finally do `spack uninstall cmake` without having to
remove all packages built with cmake.
Default package requirements might contain
variants that are not defined in each package,
so we shouldn't verify them when emitting facts
for the ASP solver.
Account for group when enforcing requirements
packages:all : don't emit facts for requirement conditions
that can't apply to current spec
#35098 added the correct extraction of toolset version for the MSVC
compiler. This updates the associated method in MSBuilder to retrieve
the (now correct) property.
This PR enables the successful execution of the spack binary cache
tutorial on Windows. It assumes gnupg and file are available (they
can be installed with choco).
* Fix handling of args with quotes in spack.bat
* `file` utility can be installed on Windows (e.g. with choco): update
error message accordingly
At least with ZSH, prefix inspections containing `./bin` result in a
`$PREFIX/./bin` and result in strange `$PATH` handling.
I.e., `module load git` will prepend `/path/to/git/./bin`, `which git`
will find the right executable, but `git --version` will print the
system one. Normalize the relative path to avoid this behavior.
See also spack/spack#31867.
fixes#34879
This commit adds a new maintainer directive,
which by default extend the list of maintainers
for a given package.
The directive is backward compatible with the current
practice of having a "maintainers" list declared at
the class level.
Move the relocation of binary text in its own class
Drop threaded text replacement, since the current bottleneck
is decompression. It would be better to parallellize over packages,
instead of over files per package.
A small improvement with separate classes for text replacement is that we
now compile the regex in the constructor; previously it was compiled per
binary to be relocated.
This commit makes explicit the format version of the spec file
we are reading from.
Before there were different functions capable of reading some
part of the spec file at multiple format versions. The decision
was implicit, since checks were based on the structure of the
JSON without ever checking a format version number.
The refactor makes also explicit which spec file format is used
by which database and lockfile format, since the information is
stored in global mappings.
To ensure we don't change the hash of old specs, JSON representations
of specs have been added as data. A unit tests checks that we read
the correct hash in, and that the hash stays the same when we
re-serialize the spec using the most recent format version.
Co-authored-by: Harmen Stoppels <me@harmenstoppels.nl>
Currently we print "sha256 checksum failed for [file]. Expected X but
got Y".
This PR extends that message with file size and contents info:
"... but got Y. File size = 123456 bytes. Contents = b'abc...def'"
That way we can immediately see if the file was downloaded only
partially, or if we downloaded a text page instead of a binary, etc.
Co-authored-by: Massimiliano Culpo <massimiliano.culpo@gmail.com>
When running unit-test the test/ci.py module is leaving
garbage (help.sh, test.sh files) in the current working
directory.
This commit changes the current working directory to a
temporary path before those files are created.
* environments: don't rewrite relative view path, expand path on cli ahead of time
Currently if you have a spack.yaml that specifies a view by relative
path, Spack expands it to an absolute path on `spack -e . install` and
persists that to disk.
This is rather annoying when you have a `spack.yaml` file inside a git
repo, cause you want to use relative paths to make it relocatable, but
you constantly have to undo the changes made to spack.yaml by Spack.
So, as an alternative:
1. Always stick to paths as they are provided in spack.yaml, never
replace them with a canonicalized version
2. Turn relative paths on the command line into absolute paths before
storing to spack.yaml. This way you can do `spack env create --dir
./env --with-view ./view` and both `./env` and `./view` are resolved
to the current working dir, as expected (not `./env/view`). This
corresponds to the old behavior of `spack env create`.
* create --with-view always takes a value
All packages with explicit Windows support can be found with
`spack list --tags=windows`.
This also removes the documentation which explicitly lists
supported packages on Windows (which is currently out of date and
is now unnecessary with the added tags).
Note that if a package does not appear in this list, it *may*
still build on Windows, but it likely means that no explicit
attempt has been made to support it.
Since SPACK_PACKAGE_IDS is now also "namespaced" with <prefix>, it makes
more sense to call the flag `--make-prefix` and alias the old flag
`--make-target-prefix` to it.
Normally when using external packages in concretization, Spack ignores
all dependencies of the external. #33777 updated this logic to attach
a Python Spec to external Python extensions (most py-* packages), but
as implemented there were a couple issues:
* this did not account for concretization groups and could generate
multiple different python specs for a single DAG
* in some cases this created a fake Python spec with insufficient
details to be usable (concretization/installation of the
extension would fail)
This PR addresses both of these issues:
* For environment specs that are concretized together, external python
extensions in those specs will all be assigned the same Python spec
* If Spack needs to "invent" a Python spec, then it will have all the
needed details (e.g. compiler/architecture)
With the new variable [prefix/]SPACK_PACKAGE_IDS you can conveniently execute
things after each successful install.
For example push just-built packages to a buildcache
```
SPACK ?= spack
export SPACK_COLOR = always
MAKEFLAGS += -Orecurse
MY_BUILDCACHE := $(CURDIR)/cache
.PHONY: all clean
all: push
ifeq (,$(filter clean,$(MAKECMDGOALS)))
include env.mk
endif
# the relevant part: push has *all* example/push/<pkg identifier> as prereqs
push: $(addprefix example/push/,$(example/SPACK_PACKAGE_IDS))
$(SPACK) -e . buildcache update-index --directory $(MY_BUILDCACHE)
$(info Pushed everything, yay!)
# and each example/push/<pkg identifier> has the install target as prereq,
# and the body can use target local $(HASH) and $(SPEC) variables to do
# things, such as pushing to a build cache
example/push/%: example/install/%
@mkdir -p $(dir $@)
$(SPACK) -e . buildcache create --allow-root --only=package --unsigned --directory $(MY_BUILDCACHE) /$(HASH) # push $(SPEC)
@touch $@
spack.lock: spack.yaml
$(SPACK) -e . concretize -f
env.mk: spack.lock
$(SPACK) -e . env depfile -o $@ --make-target-prefix example
clean:
rm -rf spack.lock env.mk example/
``
In the past we checked remote binary mirrors for existence of a spec
before attempting to download it. That changed to only checking local
copies of index.jsons (if available) to prioritize certain mirrors where
we expect to find a tarball. That was faster for CI since fetching
index.json and loading it just to order mirrors takes more time than
just attempting to fetch tarballs -- and also if we have a direct hit
there's no point to look at other mirrors.
Long story short: the info message only makes sense in the old version
of Spack, so it's better to remove it.
* Forward lookup of "test_log_file" and "test_failures"
refers #34531closes#34487fixes#34440
* Add unit test
* py-libensemble: fix tests
* Support stand-alone tests with cached files as install-time tests
Co-authored-by: Tamara Dahlgren <dahlgren1@llnl.gov>
Ensure `spack mirror add <name> <url/path>` without further arguments translates to `<name>: <url>` key value pairs in mirrors.yaml. If --s3-* flags are provided, only store the provided ones.
Co-authored-by: Harmen Stoppels <me@harmenstoppels.nl>
With this change we get the invariant that `mirror.fetch_url` and
`mirror.push_url` return valid URLs, even when the backing config
file is actually using (relative) paths with potentially `$spack` and
`$env` like variables.
Secondly it avoids expanding mirror path / URLs too early,
so if I say `spack mirror add name ./path`, it stays `./path` in my
config. When it's retrieved through MirrorCollection() we
exand it to say `file://<env dir>/path` if `./path` was set in an
environment scope.
Thirdly, the interface is simplified for the relevant buildcache
commands, so it's more like `git push`:
```
spack buildcache create [mirror] [specs...]
```
`mirror` is either a mirror name, a path, or a URL.
Resolving the relevant mirror goes as follows:
- If it contains either / or \ it is used as an anonymous mirror with
path or url.
- Otherwise, it's interpreted as a named mirror, which must exist.
This helps to guard against typos, e.g. typing `my-mirror` when there
is no such named mirror now errors with:
```
$ spack -e . buildcache create my-mirror
==> Error: no mirror named "my-mirror". Did you mean ./my-mirror?
```
instead of creating a directory in the current working directory. I
think this is reasonable, as the alternative (requiring that a local dir
exists) feels a bit pendantic in the general case -- spack is happy to
create the build cache dir when needed, saving a `mkdir`.
The old (now deprecated) format will still be available in Spack 0.20,
but is scheduled to be removed in 0.21:
```
spack buildcache create (--directory | --mirror-url | --mirror-name) [specs...]
```
This PR also touches `tmp_scope` in tests, because it didn't really
work for me, since spack fixes the possible --scope values once and
for all across tests, so tests failed when run out of order.
Sometimes I just want to know how many packages of a certain type there are.
- [x] add `--count` option to `spack list` that output the number of packages that
*would* be listed.
```console
> spack list --count
6864
> spack list --count py-
2040
> spack list --count r-
1162
```
Currently, all of the replacements in `spack.util.path.replacements()` get evaluated for
each replacement. This makes it easy to get bootstrap issues, because config is used
very early on in Spack.
Right now, if I run `test_autotools_gnuconfig_replacement_no_gnuconfig` on my M1 mac, I
get the circular reference error below. This fixes the issue by making all of the path
replacements lazy lambdas.
As a bonus, this cleans up the way we do substitution for `$env` -- it's consistent with
other substitutions now.
- [x] make all path `replacements()` lazy
- [x] clean up handling of `$env`
```console
> spack unit-test -k test_autotools_gnuconfig_replacement_no_gnuconfig
...
==> [2022-12-31-15:44:21.771459] Error: AttributeError:
The 'autotools-config-replacement' package cannot find an attribute while trying to build from sources. This might be due to a change in Spack's package format to support multiple build-systems for a single package. You can fix this by updating the build recipe, and you can also report the issue as a bug. More information at https://spack.readthedocs.io/en/latest/packaging_guide.html#installation-procedure
/Users/gamblin2/src/spack/lib/spack/spack/package_base.py:1332, in prefix:
1330 @property
1331 def prefix(self):
>> 1332 """Get the prefix into which this package should be installed."""
1333 return self.spec.prefix
Traceback (most recent call last):
File "/Users/gamblin2/src/spack/lib/spack/spack/build_environment.py", line 1030, in _setup_pkg_and_run
kwargs["env_modifications"] = setup_package(
^^^^^^^^^^^^^^
File "/Users/gamblin2/src/spack/lib/spack/spack/build_environment.py", line 757, in setup_package
set_module_variables_for_package(pkg)
File "/Users/gamblin2/src/spack/lib/spack/spack/build_environment.py", line 596, in set_module_variables_for_package
m.std_cmake_args = spack.build_systems.cmake.CMakeBuilder.std_args(pkg)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/Users/gamblin2/src/spack/lib/spack/spack/build_systems/cmake.py", line 241, in std_args
define("CMAKE_INSTALL_PREFIX", pkg.prefix),
^^^^^^^^^^
File "/Users/gamblin2/src/spack/lib/spack/spack/package_base.py", line 1333, in prefix
return self.spec.prefix
^^^^^^^^^^^^^^^^
File "/Users/gamblin2/src/spack/lib/spack/spack/spec.py", line 1710, in prefix
self.prefix = spack.store.layout.path_for_spec(self)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/Users/gamblin2/src/spack/lib/spack/spack/directory_layout.py", line 336, in path_for_spec
path = self.relative_path_for_spec(spec)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/Users/gamblin2/src/spack/lib/spack/spack/directory_layout.py", line 106, in relative_path_for_spec
projection = spack.projections.get_projection(self.projections, spec)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/Users/gamblin2/src/spack/lib/spack/spack/projections.py", line 13, in get_projection
if spec.satisfies(spec_like, strict=True):
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/Users/gamblin2/src/spack/lib/spack/spack/spec.py", line 3642, in satisfies
if not self.virtual and other.virtual:
^^^^^^^^^^^^
File "/Users/gamblin2/src/spack/lib/spack/spack/spec.py", line 1622, in virtual
return spack.repo.path.is_virtual(self.name)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/Users/gamblin2/src/spack/lib/spack/spack/repo.py", line 890, in is_virtual
return have_name and pkg_name in self.provider_index
^^^^^^^^^^^^^^^^^^^
File "/Users/gamblin2/src/spack/lib/spack/spack/repo.py", line 770, in provider_index
self._provider_index.merge(repo.provider_index)
^^^^^^^^^^^^^^^^^^^
File "/Users/gamblin2/src/spack/lib/spack/spack/repo.py", line 1096, in provider_index
return self.index["providers"]
~~~~~~~~~~^^^^^^^^^^^^^
File "/Users/gamblin2/src/spack/lib/spack/spack/repo.py", line 592, in __getitem__
self._build_all_indexes()
File "/Users/gamblin2/src/spack/lib/spack/spack/repo.py", line 607, in _build_all_indexes
self.indexes[name] = self._build_index(name, indexer)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/Users/gamblin2/src/spack/lib/spack/spack/repo.py", line 616, in _build_index
index_mtime = self.cache.mtime(cache_filename)
^^^^^^^^^^^^^^^^
File "/Users/gamblin2/src/spack/lib/spack/llnl/util/lang.py", line 826, in __getattr__
return getattr(self.instance, name)
^^^^^^^^^^^^^
File "/Users/gamblin2/src/spack/lib/spack/llnl/util/lang.py", line 825, in __getattr__
raise AttributeError()
AttributeError
```
Since we dropped support for Python 2.7, we can embrace using keyword only arguments
for many functions in Spack that use **kwargs in the function signature. Here this is done
for the llnl.util.filesystem module.
There were a couple of bugs lurking in the code related to typo-like errors when retrieving
from kwargs. Those have been fixed as well.
The code in FileCache for write_transaction attempts to delete the temporary file when an exception occurs under the context by calling shutil.rmtree. However, rmtree only operates on directories while the rest of FileCache uses normal files. This causes an empty file to be written to the cache key when unneeded.
Use os.remove instead which operates on normal files.
Co-authored-by: Harmen Stoppels <harmenstoppels@gmail.com>
Windows CMD prompt does not automatically support ASCI color control
characters on the console from Python. Enable this behavior by
accessing the current console and allowing the interpreation of ASCI
control characters from Python via the win32 API.
This commit allows (remote) spec.json files to be clearsigned and gzipped.
The idea is to reduce the number of requests and number of bytes transferred
The code in Spack to generate install and test reports currently suffers from unneeded complexity. For
instance, we have classes in Spack core packages, like `spack.reporters.CDash`, that need an
`argparse.Namespace` to be initialized and have "hard-coded" string literals on which they branch to
change their behavior:
```python
if do_fn.__name__ == "do_test" and skip_externals:
package["result"] = "skipped"
else:
package["result"] = "success"
package["stdout"] = fetch_log(pkg, do_fn, self.dir)
package["installed_from_binary_cache"] = pkg.installed_from_binary_cache
if do_fn.__name__ == "_install_task" and installed_already:
return
```
This PR attempt to polish the major issues encountered in both `spack.report` and `spack.reporters`.
Details:
- [x] `spack.reporters` is now a package that contains both the base class `Reporter` and all
the derived classes (`JUnit` and `CDash`)
- [x] Classes derived from `spack.reporters.Reporter` don't take an `argparse.Namespace` anymore
as argument to `__init__`. The rationale is that code for commands should be built upon Spack
core classes, not vice-versa.
- [x] An `argparse.Action` has been coded to create the correct `Reporter` object based on command
line arguments
- [x] The context managers to generate reports from either `spack install` or from `spack test` have
been greatly simplified, and have been made less "dynamic" in nature. In particular, the `collect_info`
class has been deleted in favor of two more specific context managers. This allows for a simpler
structure of the code, and less knowledge required to client code (in particular on which method to patch)
- [x] The `InfoCollector` class has been turned into a simple hierarchy, so to avoid conditional statements
within methods that assume a knowledge of the context in which the method is called.
On systems with remote groups, the primary user group may be remote and may not exist on
the local system (i.e., it might just be a number). On the CLI, it looks like this:
```console
> touch foo
> l foo
-rw-r--r-- 1 gamblin2 57095 0 Dec 29 22:24 foo
> chmod 2000 foo
chmod: changing permissions of 'foo': Operation not permitted
```
Here, the local machine doesn't know about per-user groups, so they appear as gids in
`ls` output. `57095` is also `gamblin2`'s uid, but the local machine doesn't know that
`gamblin2` is in the `57095` group.
Unfortunately, it seems that Python's `os.chmod()` just fails silently, setting
permissions to `0o0000` instead of `0o2000`. We can avoid this by ensuring that the file
has a group the user is known to be a member of.
- [x] Add `ensure_known_group()` in the permissions tests.
- [x] Call `ensure_known_group()` on tempfile in `test_chmod_real_entries_ignores_suid_sgid`.
There are a number of places in our docstrings where we write "list of X" as the type, even though napoleon doesn't actually support this. It ends up causing warnings when generating docs.
Now that we require Python 3, we don't have to rely on type hints in docs -- we can just use Python type hints and omit the types of args and return values from docstrings.
We should probably do this for all types in docstrings eventually, but this PR focuses on the ones that generate warnings during doc builds.
Some `mypy` annoyances we should consider in the future:
1. Adding some of these type annotations gets you:
```
note: By default the bodies of untyped functions are not checked, consider using --check-untyped-defs [annotation-unchecked]
```
because they are in unannotated functions (like constructors where we don't really need any annotations).
You can silence these with `disable_error_code = "annotation-unchecked"` in `pyproject.toml`
2. Right now we support running `mypy` in Python `3.6`. That means we have to support `mypy` `.971`, which does not support `disable_error_code = "annotation-unchecked"`, so I just filter `[annotation-unchecked]` lines out in `spack style`.
3. I would rather just turn on `check_untyped_defs` and get more `mypy` coverage everywhere, but that will require about 1,000 fixes. We should probably do that eventually.
4. We could also consider only running `mypy` on newer python versions. This is not easy to do while supporting `3.6`, because you have to use `if TYPE_CHECKING` for a lot of things to ensure that 3.6 still parses correctly. If we only supported `3.7` and above we could use [`from __future__ import annotations`](https://mypy.readthedocs.io/en/stable/runtime_troubles.html#future-annotations-import-pep-563), but we have to support 3.6 for now. Sigh.
- [x] Convert a number of docstring types to Python type hints
- [x] Get rid of "list of" wherever it appears
Local `git` tests will fail with `fatal: transport 'file' not allowed` when using git 2.38.1 or higher, due to a fix for `CVE-2022-39253`.
This was fixed in CI in #33429, but that doesn't help the issue for anyone's local environment. Instead of fixing this with git config in CI, we should ensure that the tests run anywhere.
- [x] Introduce `spack.util.git`.
- [x] Use `spack.util.git.get_git()` to get a git executable, instead of `which("git")` everywhere.
- [x] Make all `git` tests use a `git` fixture that goes through `spack.util.git.get_git()`.
- [x] Add `-c protocol.file.allow=always` to all `git` invocations under `pytest`.
- [x] Revert changes from #33429, which are no longer needed.
`spack graph` has been reworked to use:
- Jinja templates
- builder objects to construct the template context when DOT graphs are requested.
This allowed to add a new colored output for DOT graphs that highlights both
the dependency types and the nodes that are needed at runtime for a given spec.
`spack solve` is supposed to show you times you can compare. setup, ground, solve, etc.
all in a list. You're also supposed to be able to compare easily across runs. With
`pretty_seconds()` (introduced in #33900), it's easy to miss the units, e.g., spot the
bottleneck here:
```console
> spack solve --timers tcl
setup 22.125ms
load 16.083ms
ground 8.298ms
solve 848.055us
total 58.615ms
```
It's easier to see what matters if these are all in the same units, e.g.:
```
> spack solve --timers tcl
setup 0.0147s
load 0.0130s
ground 0.0078s
solve 0.0008s
total 0.0463s
```
And the units won't fluctuate from run to run as you make changes.
-[x] make `spack solve` timings consistent like before
Avoid text decoding and encoding when combining log files, instead
combine in binary mode.
Also do a buffered copy which is sometimes faster for large log files.
Currently, the Spack docs show documentation for submodules *before* documentation for
submodules on package doc pages. This means that if you put docs in `__init__.py` in
some package, the docs in there will be shown *after* the docs for all submodules of the
package instead of at the top as an intro to the package. See, e.g.,
[the lockfile docs](https://spack.readthedocs.io/en/latest/spack.environment.html#module-spack.environment),
which should be at the
[top of that page](https://spack.readthedocs.io/en/latest/spack.environment.html).
- [x] add the `--module-first` option to sphinx so that it generates module docs at top of page.
The sticky property will prevent clingo from changing the amdgpu_target
to work around conflicts. This is the same behaviour as was adopted for
cuda_arch in 055c9d125d.
Implement an alternative strategy to do index.json invalidation.
The current approach of pairs of index.json / index.json.hash is
problematic because it leads to races.
The standard solution for cache invalidation is etags, which are
supported by both http and s3 protocols, which allows one to do
conditional fetches.
This PR implements that for the http/https schemes. It should also work
for s3 schemes, but that requires other prs to be merged.
Also it improves unit tests for index.json fetches.
Interim fix for #34559
Spack's MSVC compiler definition uses ifx as the Fortran compiler.
Prior to #33385, the Spack MSVC compiler definition required the
executable to be called "ifx.exe"; #33385 replaced this with just
"ifx", which inadvertently led to ifx falsely indicating the
presence of MSVC on non-Windows systems (which leads to future
errors when attempting to query/use those compiler objects).
This commit applies a short-term fix by updating MSVC Fortran
version detection to always indicate a failure on non-Windows.
fixes#34518
Fix an issue due to the MRO chain of the package wrapper
during build. Before this PR we were always returning
False when the builder object was created before the
run_tests method was monkey patched.
This reverts commit 8035eeb36d.
And also removes logic around an additional HEAD request to prevent
a more expensive GET request on wrong content-type. Since large files
are typically an attachment and only downloaded when reading the
stream, it's not an optimization that helps much, and in fact the logic
was broken since the GET request was done unconditionally.
The main issue that's fixed is that Spack passes paths (as strings) to
functions that require urls. That wasn't an issue on unix, since there
you can simply concatenate `file://` and `path` and all is good, but on
Windows that gives invalid file urls. Also on Unix, Spack would not deal with uri encoding like x%20y for file paths.
It also removes Spack's custom url.parse function, which had its own incorrect interpretation of file urls, taking file://x/y to mean the relative path x/y instead of hostname=x and path=/y. Also it automatically interpolated variables, which is surprising for a function that parses URLs.
Instead of all sorts of ad-hoc `if windows: fix_broken_file_url` this PR
adds two helper functions around Python's own path2url and reverse.
Also fixes a bug where some `spack buildcache` commands
used `-d` as a flag to mean `--mirror-url` requiring a URL, and others
`--directory`, requiring a path. It is now the latter consistently.
When installing binary tarballs, Spack has to download from its
binary mirrors.
Sometimes Spack has cache available for these mirrors.
That cache helps to order mirrors to increase the likelihood of
getting a direct hit.
However, currently, when Spack can't find a spec in any local cache
of mirrors, it's very dumb:
- A while ago it used to query each mirror to see if it had a spec,
and use that information to order the mirror again, only to go
about and do exactly a part of what it just did: fetch the spec
from that mirror confused
- Recently, it was changed to download a full index.json, which
can be multiple dozens of MBs of data and may take a minute to
process thanks to the blazing fast performance you get with
Python.
In a typical use case of concretizing with reuse, the full index.json
is already available, and it likely that the local cache gives a perfect
mirror ordering on install. (There's typically no need to update any
caches).
However, in the use case of Gitlab CI, the build jobs don't have cache,
and it would be smart to just do direct fetches instead of all the
redundant work of (1) and/or (2).
Also, direct fetches from mirrors will soon be fast enough to
prefer these direct fetches over the excruciating slowness of
index.json files.
Writing a long dependency like:
```python
depends_on(
"llvm"
"targets=amdgpu,bpf,nvptx,webassembly"
"version_suffix=jl +link_llvm_dylib ~internal_unwind"
)
```
when it should be formatted like this:
```python
depends_on(
"llvm"
" targets=amdgpu,bpf,nvptx,webassembly"
" version_suffix=jl +link_llvm_dylib ~internal_unwind"
)
```
can cause really subtle errors. Specifically, you'll get something like this in
the package sanity tests:
```
AttributeError: 'NoneType' object has no attribute 'rpartition'
```
because Spack happily constructs a class that has a dependency with name `None`.
We can catch this earlier by banning anonymous dependency specs directly in
`depends_on()`. This causes the package itself to fail to parse, and emits
a much better error message:
```
==> Error: Invalid dependency specification in package 'julia':
llvmtargets=amdgpu,bpf,nvptx,webassemblyversion_suffix=jl +link_llvm_dylib ~internal_unwind
```
* `url_exists` improvements (take 2)
Make `url_exists` do HEAD request for http/https/s3 protocols
Rework the opener: construct it once and only once, dynamically dispatch
to the right one based on config.
It's very common for us to tell users to grep through the existing Spack packages to
find examples of what they want, and it's also very common for package developers to do
it. Now, searching packages is even easier.
`spack pkg grep` runs grep on all `package.py` files in repos known to Spack. It has no
special options other than the search string; all options passed to it are forwarded
along to `grep`.
```console
> spack pkg grep --help
usage: spack pkg grep [--help] ...
positional arguments:
grep_args arguments for grep
options:
--help show this help message and exit
```
```console
> spack pkg grep CMakePackage | head -3
/Users/gamblin2/src/spack/var/spack/repos/builtin/packages/3dtk/package.py:class _3dtk(CMakePackage):
/Users/gamblin2/src/spack/var/spack/repos/builtin/packages/abseil-cpp/package.py:class AbseilCpp(CMakePackage):
/Users/gamblin2/src/spack/var/spack/repos/builtin/packages/accfft/package.py:class Accfft(CMakePackage, CudaPackage):
```
```console
> spack pkg grep -Eho '(\S*)\(PythonPackage\)' | head -3
AwsParallelcluster(PythonPackage)
Awscli(PythonPackage)
Bueno(PythonPackage)
```
## Return Value
This retains the return value semantics of `grep`:
* 0 for found,
* 1 for not found
* >1 for error
## Choosing a `grep`
You can set the ``SPACK_GREP`` environment variable to choose the ``grep``
executable this command should use.
Unit tests on Windows are supposed to pass for any PR to pass CI.
However, the return code for the unit test command was not being
checked, which meant this check was always passing (effectively
disabled). This PR
* Properly checks the result of the unit tests and fails if the
unit tests fail
* Fixes (or disables on Windows) a number of tests which have
"drifted" out of support on Windows since this check was
effectively disabled
## Motivation
Our parser grew to be quite complex, with a 2-state lexer and logic in the parser
that has up to 5 levels of nested conditionals. In the future, to turn compilers into
proper dependencies, we'll have to increase the complexity further as we foresee
the need to add:
1. Edge attributes
2. Spec nesting
to the spec syntax (see https://github.com/spack/seps/pull/5 for an initial discussion of
those changes). The main attempt here is thus to _simplify the existing code_ before
we start extending it later. We try to do that by adopting a different token granularity,
and by using more complex regexes for tokenization. This allow us to a have a "flatter"
encoding for the parser. i.e., it has fewer nested conditionals and a near-trivial lexer.
There are places, namely in `VERSION`, where we have to use negative lookahead judiciously
to avoid ambiguity. Specifically, this parse is ambiguous without `(?!\s*=)` in `VERSION_RANGE`
and an extra final `\b` in `VERSION`:
```
@ 1.2.3 : develop # This is a version range 1.2.3:develop
@ 1.2.3 : develop=foo # This is a version range 1.2.3: followed by a key-value pair
```
## Differences with the previous parser
~There are currently 2 known differences with the previous parser, which have been added on purpose:~
- ~No spaces allowed after a sigil (e.g. `foo @ 1.2.3` is invalid while `foo @1.2.3` is valid)~
- ~`/<hash> @1.2.3` can be parsed as a concrete spec followed by an anonymous spec (before was invalid)~
~We can recover the previous behavior on both ones but, especially for the second one, it seems the current behavior in the PR is more consistent.~
The parser is currently 100% backward compatible.
## Error handling
Being based on more complex regexes, we can possibly improve error
handling by adding regexes for common issues and hint users on that.
I'll leave that for a following PR, but there's a stub for this approach in the PR.
## Performance
To be sure we don't add any performance penalty with this new encoding, I measured:
```console
$ spack python -m timeit -s "import spack.spec" -c "spack.spec.Spec(<spec>)"
```
for different specs on my machine:
* **Spack:** 0.20.0.dev0 (c9db4e50ba045f5697816187accaf2451cb1aae7)
* **Python:** 3.8.10
* **Platform:** linux-ubuntu20.04-icelake
* **Concretizer:** clingo
results are:
| Spec | develop | this PR |
| ------------- | ------------- | ------- |
| `trilinos` | 28.9 usec | 13.1 usec |
| `trilinos @1.2.10:1.4.20,2.0.1` | 131 usec | 120 usec |
| `trilinos %gcc` | 44.9 usec | 20.9 usec |
| `trilinos +foo` | 44.1 usec | 21.3 usec |
| `trilinos foo=bar` | 59.5 usec | 25.6 usec |
| `trilinos foo=bar ^ mpich foo=baz` | 120 usec | 82.1 usec |
so this new parser seems to be consistently faster than the previous one.
## Modifications
In this PR we just substituted the Spec parser, which means:
- [x] Deleted in `spec.py` the `SpecParser` and `SpecLexer` classes. deleted `spack/parse.py`
- [x] Added a new parser in `spack/parser.py`
- [x] Hooked the new parser in all the places the previous one was used
- [x] Adapted unit tests in `test/spec_syntax.py`
## Possible future improvements
Random thoughts while working on the PR:
- Currently we transform hashes and files into specs during parsing. I think
we might want to introduce an additional step and parse special objects like
a `FileSpec` etc. in-between parsing and concretization.
This commit reworks the bootstrapping procedure to use Spack environments
as much as possible.
The `spack.bootstrap` module has also been reorganized into a Python package.
A distinction is made among "core" Spack dependencies (clingo, GnuPG, patchelf)
and other dependencies. For a number of reasons, explained in the `spack.bootstrap.core`
module docstring, "core" dependencies are bootstrapped with the current ad-hoc
method.
All the other dependencies are instead bootstrapped using a Spack environment
that lives in a directory specific to the interpreter and the architecture being used.
All the vermin annotations we were using were for optional features introduced in early
Python 3 versions. We no longer need any of them, as we only support Python 3.6+. If we
start optionally using features from newer Pythons than 3.6, we will need more vermin
annotations.
Co-authored-by: Harmen Stoppels <harmenstoppels@gmail.com>
Co-authored-by: Harmen Stoppels <harmenstoppels@gmail.com>
We no longer support Python <3.6, so we don't need to check whether Python supports SSL
verification in `spack.util.web`.
- [x] Remove a bunch of logic we needed to appease Python 2
We've stopped supporting Python 2, and contributors are noticing that our CI no longer
allows Python 2.7 comment type hints. They end up having to adapt them, but this adds
extra unrelated work to PRs.
- [x] Move to 3.6 type hints across the entire code base
All Spec attributes are now represented as `attr(attribute_name, ... args ...)`, e.g.
`attr(node, "hdf5")` instead of `node("hdf5")`, as we *have* to maintain the `attr()`
form anyway, and it simplifies the encoding to just maintain one form of the Spec
information.
Background
----------
In #20644, we unified the way conditionals are done in the concretizer, but this
introduced a nasty aspect to the encoding: we have to maintain everything we want in
general conditions in two forms: `predicate(...)` and `attr("predicate", ...)`. For
example, here's the start of the table of spec attributes we had to maintain:
```prolog
node(Package) :- attr("node", Package).
virtual_node(Virtual) :- attr("virtual_node", Virtual).
hash(Package, Hash) :- attr("hash", Package, Hash).
version(Package, Version) :- attr("version", Package, Version).
...
```
```prolog
attr("node", Package) :- node(Package).
attr("virtual_node", Virtual) :- virtual_node(Virtual).
attr("hash", Package, Hash) :- hash(Package, Hash).
attr("version", Package, Version) :- version(Package, Version).
...
```
This adds cognitive load to understanding how the concretizer works, as you have to
understand the equivalence between the two forms of spec attributes. It also makes the
general condition logic in #20644 hard to explain, and it's easy to forget to add a new
equivalence to this list when adding new spec attributes (at least two people have been
bitten by this).
Solution
--------
- [x] remove the equivalence list from `concretize.lp`
- [x] simplify `spec_clauses()`, `condition()`, and other functions in `asp.py` that need
to deal with `Spec` attributes.
- [x] Convert all old-form spec attributes in `concretize.lp` to the `attr()` form
- [x] Simplify `display.lp`, where we also had to maintain a list of spec attributes. Now
we only need to show `attr/2`, `attr/3`, and `attr/4`.
- [x] Simplify model extraction logic in `asp.py`.
Performance
-----------
This seems to result in a smaller grounded problem (as there are no longer duplicated
`attr("foo", ...)` / `foo(...)` predicates in the program), but it also adds a slight
performance overhead vs. develop. Ultimately, simplifying the encoding will be a win,
particularly for improving error messages.
Notes
-----
This will simplify future node refactors in `concretize.lp` (e.g., not identifying nodes
by package name, which we need for separate build dependencies).
I'm still not entirely used to reading `attr()` notation, but I thnk it's ultimately
clearer than what we did before. We need more uniform naming, and it's now clear what is
part of a solution. We should probably continue making the encoding of `concretize.lp`
simpler and more self-explanatory. It may make sense to rename `attr` to something like
`node_attr` and to simplify the names of node attributes. It also might make sense to do
something similar for other types of predicates in `concretize.lp`.
This reverts commit d06fd26c9a.
The problem is that Bitbucket's API forwards download requests to an S3 bucket using a temporary URL. This URL includes a signature for the request, which embeds the HTTP verb. That means only GET requests are allowed, and HEAD requests would fail verification, leading to 403 erros. The same is observed when using `curl -LI ...`
Using `-Werror` is good practice for development and testing, but causes us a great
deal of heartburn supporting multiple compiler versions, especially as newer compiler
versions add warnings for released packages. This PR adds support for suppressing
`-Werror` through spack's compiler wrappers. There are currently three modes for
the `flags:keep_werror` setting:
* `none`: (default) cancel all `-Werror`, `-Werror=*` and `-Werror-*` flags by
converting them to `-Wno-error[=]*` flags
* `specific`: preserve explicitly selected warnings as errors, such as
`-Werror=format-truncation`, but reverse the blanket `-Werror`
* `all`: keeps all `-Werror` flags
These can be set globally in config.yaml, through the config command-line flags, or
overridden by a particular package (some packages use Werror as a proxy for determining
support for other compiler features). We chose to use this approach because:
1. removing `-Werror` flags entirely broke *many* build systems, especially autoconf
based ones, because of things like checking `-Werror=feature` and making the
assumption that if that did not error other flags related to that feature would also work
2. Attempting to preserve `-Werror` in some phases but not others caused similar issues
3. The per-package setting came about because some packages, even with all these
protections, still use `-Werror` unsafely. Currently there are roughly 3 such packages
known.
For reasons beyond me Python thinks it's a great idea to upgrade HEAD
requests to GET requests when following redirects. So, this PR adds a
better `HTTPRedirectHandler`, and also moves some ad-hoc logic around
for dealing with disabling SSL certs verification.
Also, I'm stumped by the fact that Spack's `url_exists` does not use
HEAD requests at all, so in certain cases Spack awkwardly downloads
something first to see if it can download it, and then downloads it
again because it knows it can download it. So, this PR ensures that both
urllib and botocore use HEAD requests.
Finally, it also removes some things that were there to support currently
unsupported Python versions.
Notice that the HTTP spec [section 10.3.2](https://datatracker.ietf.org/doc/html/rfc2616.html#section-10.3.2) just talks about how to deal
with POST request on redirect (whether to follow or not):
> If the 301 status code is received in response to a request other
> than GET or HEAD, the user agent MUST NOT automatically redirect the
> request unless it can be confirmed by the user, since this might
> change the conditions under which the request was issued.
> Note: When automatically redirecting a POST request after
> receiving a 301 status code, some existing HTTP/1.0 user agents
> will erroneously change it into a GET request.
Python has a comment about this, they choose to go with the "erroneous change".
But they then mess up the HEAD request while following the redirect, probably
because they were too busy discussing how to deal with POST.
See https://github.com/python/cpython/pull/99731
This adds super-lazy maintainer mode to `spack checksum`: Instead of
only printing the new checksums to the terminal, `-a` and
`--add-to-package` will add the new checksums to the `package.py` file
and open it in the editor afterwards for final checks.
* Add a WindowsRegistryView class, which can query for existing
package installations on Windows. This is particularly important
because some Windows packages (including those added here)
do not allow two simultaneous installs, and this can be
queried in order to provide a clear error message.
* Consolidate external path detection logic for Windows into
WindowsKitExternalPaths and WindowsCompilerExternalPaths objects.
* Add external-only packages win-sdk and wgl
* Add win-wdk (including external detection) which depends on
win-sdk
* Replace prior msmpi implementation with a source-based install
(depends on win-wdk). This install can control the install
destination (unlike the binary installation).
* Update MSVC compiler to choose vcvars based on win-sdk dependency
* Provide "msbuild" module-level variable to packages during build
* When creating symlinks on Windows, need to explicitly specify when
a symlink target is a directory
* executables_in_path no-longer defaults to using PATH (this is
now expected to be taken care of by the caller)
Spec traversals can now specify a topological ordering. A topologically-
ordered traversal with input specs X1, X2... will
* include all of X1, X2... and their children
* be ordered such that a given node is guaranteed to appear before any
of its children in the traversal
Other notes:
* Input specs can be children of other input specs (this is useful if
a user specifies a set of specs to uninstall: some of those specs
might be children of others)
* `direction="parents"` will produce a reversed topological order
(children always come before parents).
* `cover="edges"` will generate a list of edges L such that (a) input
edges will always appear before output edges and (b) if you create
a list with the destination of each edge in L the result is
topologically ordered
* test_suite.py: speed up slow test by using mock packages
* Don't resolve the sha during unit-tests
* Skip long-running test that fails, instead of executing it
* uninstall: fix accidental cubic complexity
Currently spack uninstall runs in worst case cubic time complexity
thanks to traversal during traversal during traversal while collecting
the specs to be uninstalled.
Also brings down the number of error messages printed to something
linear in the amount of matching specs instead of quadratic.
* Add a regression test for 33928
* PackageBase should not set `(build|install)_time_test_callbacks`
* Fix audits by preserving the current semantic
Co-authored-by: Massimiliano Culpo <massimiliano.culpo@gmail.com>
* Enable hdf5 build (including +mpi) on Windows
* This includes updates to hdf5 dependencies openssl (minor edit) and
bzip2 (more-extensive edits)
* Add binary-based installation of msmpi (this is currently the only
supported MPI implementation in Spack for Windows). Note that this
does not install to the Spack-specified prefix. This implementation
will be replaced with a source-based implementation
Co-authored-by: John Parent <john.parent@kitware.com>
#32942 fixed bootstrapping on Windows by having the core Spack
code explicitly add the Clingo package bin/ directory as a
DLL path.
Since then, #33400 has been merged, which ensures that the Python
module installed by the Spack `clingo` package can find the DLLs
in bin/.
Note that this only works for Spack instances which have been
bootstrapped after #33400: for installations bootstrapped before
then, you will need to run `spack clean -b` (this would only
be needed for Spack instances running on Windows).
Revamp the timer so we always have a designated begin and end.
Fix a bug where the phase timer was stopped before the phase started,
resulting in incorrect timing reports in timers.json.
Add spack.ld_so_conf.host_dynamic_linker_search_paths
Retrieve the current host runtime search paths for shared libraries;
for GNU and musl Linux we try to retrieve the dynamic linker from the
current Python interpreter and then find the corresponding config file
(e.g. ld.so.conf or ld-musl-<arch>.path). Similar can be done for
BSD and others, but this is not implemented yet. The default paths
are always returned. We don't check if the listed directories exist.
Use this in spack external find for libraries.
Co-authored-by: Harmen Stoppels <harmenstoppels@gmail.com>
While binaries built for PRs that get merged must still be rebuilt
in develop pipelines, they can be used by other PRs that find they
would otherwise need to rebuild them. Now that spackbot is
managing copying PR binaries from merged PRs into a shared location,
keeping it pruned to a reasonable size, and making sure the indices
are up to date, spack can use these mirrors as a potential source
of binaries.
I'm finding I often want the date in my paths and it would be nice if spack had a config variable for this.
Co-authored-by: Tamara Dahlgren <35777542+tldahlgren@users.noreply.github.com>
* Remove CI jobs related to Python 2.7
* Remove Python 2.7 specific code from Spack core
* Remove externals for Python 2 only
* Remove llnl.util.compat
We added a hotfix to releases/v0.19 with a feature flag, but the flag
is incompatible with the config schema on `develop`.
- [x] Ensure schema is compatible on develop even though config option is unused.
* Speed-up bootstrap mirror unit test
The unit test doesn't need to concretize, since it checks
only metadata for the mirror.
* architecture.py: use "default_mock_concretization" for slow test
Environments and environment views have taken over the role of `spack activate/deactivate`, and we should deprecate these commands for several reasons:
- Global activation is a really poor idea:
- Install prefixes should be immutable; since they can have multiple, unrelated dependents; see below
- Added complexity elsewhere: verification of installations, tarballs for build caches, creation of environment views of packages with unrelated extensions "globally activated"... by removing the feature, it gets easier for people to contribute, and we'd end up with fewer bugs due to edge cases.
- Environment accomplish the same thing for non-global "activation" i.e. `spack view`, but better.
Also we write in the docs:
```
However, Spack global activations have two potential drawbacks:
#. Activated packages that involve compiled C extensions may still
need their dependencies to be loaded manually. For example,
``spack load openblas`` might be required to make ``py-numpy``
work.
#. Global activations "break" a core feature of Spack, which is that
multiple versions of a package can co-exist side-by-side. For example,
suppose you wish to run a Python package in two different
environments but the same basic Python --- one with
``py-numpy@1.7`` and one with ``py-numpy@1.8``. Spack extensions
will not support this potential debugging use case.
```
Now that environments are established and views can take over the role of activation
non-destructively, we can remove global activation/deactivation.
Currently, external `PythonPackage`s cause install failures because the logic in `PythonPackage` assumes that it can ask for `spec["python"]`. Because we chop off externals' dependencies, an external Python extension may not have a `python` dependency.
This PR resolves the issue by guaranteeing that a `python` node is present in one of two ways:
1. If there is already a `python` node in the DAG, we wire the external up to it.
2. If there is no existing `python` node, we wire up a synthetic external `python` node, and we assume that it has the same prefix as the external.
The assumption in (2) isn't always valid, but it's better than leaving the user with a non-working `PythonPackage`.
The logic here is specific to `python`, but other types of extensions could take advantage of it. Packages need only define `update_external_dependencies(self)`, and this method will be called on externals after concretization. This likely needs to be fleshed out in the future so that any added nodes are included in concretization, but for now we only bolt on dependencies post-concretization.
Co-authored-by: Todd Gamblin <tgamblin@llnl.gov>
Spack currently creates a temporary sbang that is moved "atomically" in place,
but this temporary causes races when multiple processes start installing sbang.
Let's just stick to an idempotent approach. Notice that we only re-install sbang
if Spack updates it (since we do file compare), and sbang was only touched
18 times in the past 6 years, whereas we hit the sbang tempfile issue
frequently with parallel install on a fresh spack instance in CI.
Also fixes a bug where permissions weren't updated if config changed but
the latest version of the sbang file was already installed.
The `intel` compiler at versions > 20 is provided by the `intel-oneapi-compilers-classic`
package (a thin wrapper around the `intel-oneapi-compilers` package), and the `oneapi`
compiler is provided by the `intel-oneapi-compilers` package.
Prior to this work, neither of these compilers could be bootstrapped by Spack as part of
an install with `install_missing_compilers: True`.
Changes made to make these two packages bootstrappable:
1. The `intel-oneapi-compilers-classic` package includes a bin directory and symlinks
to the compiler executables, not just logical pointers in Spack.
2. Spack can look for bootstrapped compilers in directories other than `$prefix/bin`,
defined on a per-package basis
3. `intel-oneapi-compilers` specifies a non-default search directory for the
compiler executables.
4. The `spack.compilers` module now can make more advanced associations between
packages and compilers, not just simple name translations
5. Spack support for lmod hierarchies accounts for differences between package
names and the associated compiler names for `intel-oneapi-compilers/oneapi`,
`intel-oneapi-compilers-classic/intel@20:`, `llvm+clang/clang`, and
`llvm-amdgpu/rocmcc`.
- [x] full end-to-end testing
- [x] add unit tests
"spack install foo" no longer adds package "foo" to the environment
(i.e. to the list of root specs) by default: you must specify "--add".
Likewise "spack uninstall foo" no longer removes package "foo" from
the environment: you must specify --remove. Generally this means
that install/uninstall commands will no longer modify the users list
of root specs (which many users found problematic: they had to
deactivate an environment if they wanted to uninstall a spec without
changing their spack.yaml description).
In more detail: if you have environments e1 and e2, and specs [P, Q, R]
such that P depends on R, Q depends on R, [P, R] are in e1, and [Q, R]
are in e2:
* `spack uninstall --dependents --remove r` in e1: removes R from e1
(but does not uninstall it) and uninstalls (and removes) P
* `spack uninstall -f --dependents r` in e1: will uninstall P, Q, and
R (i.e. e2 will have dependent specs uninstalled as a side effect)
* `spack uninstall -f --dependents --remove r` in e1: this uninstalls
P, Q, and R, and removes [P, R] from e1
* `spack uninstall -f --remove r` in e1: uninstalls R (so it is
"missing" in both environments) and removes R from e1 (note that e1
would still install R as a dependency of P, but it would no longer
be listed as a root spec)
* `spack uninstall --dependents r` in e1: will fail because e2 needs R
Individual unit tests were created for each of these scenarios.
Somehow a network error when cloning the repo for ci gets
categorized by gitlab as a script failure. To make sure we retry
jobs that failed for that reason or a similar one, include
"script_failure" as one of the reasons for retrying service jobs
(which include "no specs to rebuild" jobs, update buildcache
index jobs, and temp storage cleanup jobs.
Add a `project` block to the toml config along with development and CI
dependencies and a minimal `build-system` block, doing basically
nothing, so that spack can be bootstrapped to a full development
environment with:
```shell
$ hatch -e dev shell
```
or for a minimal environment without hatch:
```shell
$ python3 -m venv venv
$ source venv/bin/activate
$ python3 -m pip install --upgrade pip
$ python3 -m pip install -e '.[dev]'
```
This means we can re-use the requirements list throughout the workflow
yaml files and otherwise maintain this list in *one place* rather than
several disparate ones. We may be stuck with a couple more temporarily
to continue supporting python2.7, but aside from that it's less places
to get out of sync and a couple new bootstrap options.
Co-authored-by: Adam J. Stewart <ajstewart426@gmail.com>
This change uses the aws cli, if available, to retrieve spec files
from the mirror to a local temp directory, then parallelizes the
reading of those files from disk using multiprocessing.ThreadPool.
If the aws cli is not available, then a ThreadPool is used to fetch
and read the spec files from the mirror.
Using aws cli results in ~16 times speed up to recreate the binary
mirror index, while just parallelizing the fetching and reading
results in ~3 speed up.
The compiler bootstrapping logic currently does not add a task when the compiler package is already in the install task queue. This causes failures when the compiler package is added without the additional metadata telling the task to update the compilers list.
Solution: requeue compilers for bootstrapping when needed, to update `task.compiler` metadata.
Currently, develop specs that are not roots and are not explicitly listed dependencies
of the roots are not applied.
- [x] ensure dev specs are applied.
Co-authored-by: Todd Gamblin <tgamblin@llnl.gov>
`spack env create` enables a view by default (in a weird hidden
directory, but well...). This is asking for trouble with the other
default of `concretizer:unify:false`, since having different flavors of
the same spec in an environment, leads to collision errors when
generating the view.
A change of defaults would improve user experience:
However, `unify:true` makes most sense, since any time the issue is
brought up in Slack, the user changes the concretization config, since
it wasn't the intention to have different flavors of the same spec, and
install times are decreased.
Further we improve the docs and drop the duplicate root spec limitation
Dependencies specified by hash are unique in Spack in that the abstract
specs are created with internal structure. In this case, the constraint
generation for spec matrices fails due to flattening the structure.
It turns out that the dep_difference method for Spec.constrain does not
need to operate on transitive deps to ensure correctness. Removing transitive
deps from this method resolves the bug.
- [x] Includes regression test
Without this, Meson will use its Wraps to automatically download and
install dependencies. We want to manage dependencies explicitly,
therefore disable this functionality.
Currently, Spack can fail for a valid spec if the spec is constructed from overlapping, but not conflicting, concrete specs via the hash.
For example, if abcdef and ghijkl are the hashes of specs that both depend on zlib/mnopqr, then foo ^/abcdef ^/ghijkl will fail to construct a spec, with the error message "Cannot depend on zlib... twice".
This PR changes this behavior to check whether the specs are compatible before failing.
With this PR, foo ^/abcdef ^/ghijkl will concretize.
As a side-effect, so will foo ^zlib ^zlib and other specs that are redundant on their dependencies.
Argparse started raising ArgumentError exceptions
when the same parser is added twice. Therefore, we
perform the addition only if the parser is not there
already
Port match syntax to our unparser
Compilers and linker optimize string constants for space by aliasing
them when one is a suffix of another. For gcc / binutils this happens
already at -O1, due to -fmerge-constants. This means that we have
to take care during relocation to always preserve a certain length
of the suffix of those prefixes that are C-strings.
In this commit we pick length 7 as a safe suffix length, assuming the
suffix is typically the 7 characters from the hash (i.e. random), so
it's unlikely to alias with any string constant used in the sources.
In general we now pad shortened strings from the left with leading
dir seperators, but in the case of C-strings that are much shorter
and don't share a common suffix (due to projections), we do allow
shrinking the C-string, appending a null, and retaining the old part
of the prefix.
Also when rewiring, we ensure that the new hash preserves the last
7 bytes of the old hash.
Co-authored-by: Harmen Stoppels <harmenstoppels@gmail.com>
A user may want to set some attributes on a package without actually modifying the package (e.g. if they want to git pull updates to the package without conflicts). This PR adds a per-package configuration section called "set", which is a dictionary of attribute names to desired values. For example:
packages:
openblas:
package_attributes:
submodules: true
git: "https://github.com/myfork/openblas"
in this case, the package will always retrieve git submodules, and will use an alternate location for the git repo.
While git, url, and submodules are the attributes for which we envision the most usage, this allows any attribute to be overridden, and the acceptable values are any value parseable from yaml.
Newer versions of the CrayPE for EX systems have standalone compiler executables for CCE and compiler wrappers for Cray MPICH. With those, we can treat the cray systems as part of the linux platform rather than having a separate cray platform.
This PR:
- [x] Changes cray platform detection to ignore EX systems with Craype version 21.10 or later
- [x] Changes the cce compiler to be detectable via paths
- [x] Changes the spack compiler wrapper to understand the executable names for the standalone cce compiler (`craycc`, `crayCC`, `crayftn`).
Whenever the rpath string actually _grows_, it falls back to patchelf,
when it stays the same length or gets shorter, we update it in-place,
padded with null bytes.
This PR only deals with absolute -> absolute rpath replacement. We don't
use `_build_tarball(relative=True)` in our CI. If `relative` then it falls
back to the old replacement code.
With this PR, relocation time goes down significantly, likely because patchelf
does some odd things with mmap, causing lots of overhead. Example:
- `binutils`: 700MB installed, goes from `1.91s` to `0.57s`, or `3.4x` faster.
Relocation time: 27% -> 10% of total install time
- `llvm`: 6.8GB installed, goes from `28.56s` to `5.38`, or `5.3x` faster.
Relocation time: 44% -> 13% of total install time
The bottleneck is now decompression.
Note: I'm somewhat confused about the "relative rpath" code paths. Right
now this PR only deals with absolute -> absolute replacement. As far as
I understand, if you embrace relative rpaths when uploading to the
buildcache, the whole point is you _don't_ want to patch rpaths on
install? So it seems fine to not expand `$ORIGIN` again imho.
When a package asks for non-parallel make, we need to force `make -j1` because just doing `make` will run in parallel under jobserver (e.g. `spack env depfile`).
We now always add `-j1` when asked for a non-parallel execution (even if there is no jobserver).
And each `MakeExecutable` can now ask for jobserver support or not. For example: the default `ninja` does not support jobserver so spack applies the default `-j`, but `ninja@kitware` or `ninja-fortran` does, so spack doesn't add `-j`.
Tips: you can run `SPACK_INSTALL_FLAGS=-j1 make -f spack-env-depfile.make -j8` to avoid massive job-spawning because of build tools that don't support jobserver (ninja).
We try to avoid non-default variant values in the concretizer, but this doesn't make
sense for variants forced to take some non-default value by variant propagation.
Counting this as a penalty effectively biases the concretizer for small specs dependency
graphs -- something we try very hard to avoid elsewhere because it can lead to very
strange decisions.
Example: with the penalty, `spack spec hdf5` will choose the default `openmpi` as its
`mpi` provider, but `spack spec hdf5 ~~shared` will choose `mpich` because it has to set
fewer non-default variant values because `mpich`'s DAG is smaller. That's not a good
reason to prefer a non-default virtual provider.
To fix this, if the user explicitly requests a non-default value to be propagated, there
shouldn't be a penalty. Variant values set on the CLI already don't count as default; we
just need to extend that to propagated values.
Adds another post install hook that loops over the install prefix, looking for shared libraries type of ELF files, and sets the soname to their own absolute paths.
The idea being, whenever somebody links against those libraries, the linker copies the soname (which is the absolute path to the library) as a "needed" library, so that at runtime the dynamic loader realizes the needed library is a path which should be loaded directly without searching.
As a result:
1. rpaths are not used for the fixed/static list of needed libraries in the dynamic section (only for _actually_ dynamically loaded libraries through `dlopen`), which largely solves the issue that Spack's rpaths are a heuristic (`<prefix>/lib` and `<prefix>/lib64` might not be where libraries really are...)
2. improved startup times (no library search required)
Untouched spec pruning was added to reduce the number of specs
developers see getting rebuilt in their PR pipelines that they
don't understand. Because the state of the develop mirror lags
quite far behind the tip of the develop branch, PRs often find
they need to rebuild things untouched by their PR.
Untouched spec pruning was previously implemented by finding all
specs in the environment with names of packages touched by the PR,
traversing in both directions the DAGS of those specs, and adding
all dependencies as well as dependents to a list of concrete specs
that should not be considered for pruning.
We found that this heuristic results in too many pruned specs, and
that dependents of touched specs must have all their dependencies
added to the list of specs that should not be considered for pruning.
This issue was introduced in #29761:
```
==> Installing ncurses-6.3-22hz6q6cvo3ep2uhrs3erpp2kogxncbn
==> No binary for ncurses-6.3-22hz6q6cvo3ep2uhrs3erpp2kogxncbn found: installing from source
==> Using cached archive: /spack/var/spack/cache/_source-cache/archive/97/97fc51ac2b085d4cde31ef4d2c3122c21abc217e9090a43a30fc5ec21684e059.tar.gz
==> No patches needed for ncurses
==> ncurses: Executing phase: 'autoreconf'
==> ncurses: Executing phase: 'configure'
==> ncurses: Executing phase: 'build'
==> ncurses: Executing phase: 'install'
==> Error: AttributeError: 'str' object has no attribute 'propagate'
The 'ncurses' package cannot find an attribute while trying to build from sources. This might be due to a change in Spack's package format to support multiple build-systems for a single package. You can fix this by updating the build recipe, and you can also report the issue as a bug. More information at https://spack.readthedocs.io/en/latest/packaging_guide.html#installation-procedure
/spack/lib/spack/spack/build_environment.py:1075, in _setup_pkg_and_run:
1072 tb_string = traceback.format_exc()
1073
1074 # build up some context from the offending package so we can
>> 1075 # show that, too.
1076 package_context = get_package_context(tb)
1077
1078 logfile = None
```
It turns out this was caused by a bug that had been around much longer, in which the flags were passed by reference to the flag_handler, and the flag_handler was modifying the spec object, not just the flags given to the build system. The scope of this bug was limited by the forking model in Spack, which is how it went under the radar for so long.
PR includes regression test.
* remove deptype_query remnants
* deptypes -> deptype
These arguments haven't existed since 2017, but `traverse` now fails on unknown **kwargs, so they have finally popped up.
This updates the propagation logic used in `concretize.lp` to avoid rules with `path()`
in the body and instead base propagation around `depends_on()`.
Currently, compiler flags and variants are inconsistent: compiler flags set for a
package are inherited by its dependencies, while variants are not. We should have these
be consistent by allowing for inheritance to be enabled or disabled for both variants
and compiler flags.
- [x] Make new (spec language) operators
- [x] Apply operators to variants and compiler flags
- [x] Conflicts currently result in an unsatisfiable spec
(i.e., you can't propagate two conflicting values)
What I propose is using two of the currently used sigils to symbolized that the variant
or compiler flag will be inherited:
Example syntax:
- `package ++variant`
enabled variant that will be propagated to dependencies
- `package +variant`
enabled variant that will NOT be propagated to dependencies
- `package ~~variant`
disabled variant that will be propagated to dependencies
- `package ~variant`
disabled variant that will NOT be propagated to dependencies
- `package cflags==True`
`cflags` will be propagated to dependencies
- `package cflags=True`
`cflags` will NOT be propagated to dependencies
Syntax for string-valued variants is similar to compiler flags.
Fixes an issue on the RHEL8 UBI container where this test would fail because `gr_mem`
was empty for every entry in the `grp` DB.
You have to check *both* the `pwd` database (which has primary groups) and `grp` (which
has other gorups) to do this correctly.
- [x] update `llnl.util.filesystem.group_ids()` to do this
- [x] use it in the `sbang` test
This PR introduces breadth-first traversal, and moves depth-first traversal
logic out of Spec's member functions, into `traverse.py`.
It introduces a high-level API with three main methods:
```python
spack.traverse.traverse_edges(specs, kwargs...)
spack.traverse.traverse_nodes(specs, kwags...)
spack.traverse.traverse_tree(specs, kwargs...)
```
with the usual `root`, `order`, `cover`, `direction`, `deptype`, `depth`, `key`,
`visited` kwargs for the first two.
What's new is that `order="breadth"` is added for breadth-first traversal.
The lower level API is not exported, but is certainly useful for advanced use
cases. The lower level API includes visitor classes for direction reversal and
edge pruning, which can be used to create more advanced traversal methods,
especially useful when the `deptype` is not constant but depends on the node
or depth.
---
There's a couple nice use-cases for breadth-first traversal:
- Sometimes roots have to be handled differently (e.g. follow build edges of
roots but not of deps). BFS ensures that root nodes are always discovered at
depth 0, instead of at any depth > 1 as a dep of another root.
- When printing a tree, it would be nice to reduce indent levels so it fits in the
terminal, and ensure that e.g. `zlib` is not printed at indent level 10 as a
dependency of a build dep of a build dep -- rather if it's a direct dep of my
package, I wanna see it at depth 1. This basically requires one breadth-first
traversal to construct a tree, which can then be printed with depth-first traversal.
- In environments in general, it's sometimes inconvenient to have a double
loop: first over the roots then over each root's deps, and maintain your own
`visited` set outside. With BFS, you can simply init the queue with the
environment root specs and it Just Works. [Example here](3ec7304699/lib/spack/spack/environment/environment.py (L1815-L1816))
Currently, many tests hardcode to older versions of gcc for comparisons of
concretization among compiler versions. Those versions are too old to concretize for
`aarch64`-family targets, which leads to failing tests on `aarch64`.
This PR fixes those tests by updating the compiler versions used for testing.
Currently, many tests hardcode the expected architecture result in concretization to the
`x86_64` family of architectures.
This PR generalizes the tests that can be generalized, to cover multiple architecture
families. For those that test specific relationships among `x86_64`-family targets, it
ensures that concretization uses the `x86_64`-family targets in those cases.
Currently, many tests rely on the fact that `AutotoolsPackage` imposes no dependencies
on the inheriting package. That is not true on `aarch64`-family architectures.
This PR ensures that the fact `AutotoolsPackage` on `aarch64` pulls in a dependency on
`gnuconfig` is ignored when testing for the appropriate relationships among dependencies
Additionally, 5 tests currently prompt the user for input when `gpg` is available in the
user's path. This PR fixes that issue. And 7 tests fail currently when the user has a
yubikey available. This PR fixes the incorrect gpg argument causing those issues.
The `spack info <package>` command does not show the `Virtual Packages:` output unless the `--virtuals` command option is passed. Before this changes, the information that the command is supposed to be illustrating is not shown in the example and is confusing.
Changes to improve locating shared libraries on Windows, which in
turn enables the use of Clingo. This PR attempts to establish a
proper distinction between linking on Windows vs. Linux/Mac: on
Windows, linking is always done with .lib files (never .dll files).
This somewhat complicates the model since the Spec.lib method could
return libraries that were used for both linking and loading, but
since these are not always the same on Windows, it was decided to
treat Spec.libs as being for link-time libraries. Additional functions
are added to help dependents locate run-time libraries.
* Clingo is now the default concretizer on Windows
* Clingo is now the concretizer used for unit tests on Windows
* Fix a permissions issue that can occur while moving Git files during
fetching/staging
* Packages can now implement "win_add_library_dependent" to register
files/directories that include libraries that would need to link
to dependency dlls
* Packages can now implement "win_add_rpath" to register the locations
of dlls that dependents would want to load
* "Spec.libs" on Windows is updated to return link-time libraries
(i.e. .lib files, rather than .dll files)
* PackageBase.rpath on Windows is now updated to return the most-likely
locations where .dlls will be found (which is generally in the bin/
directory)
Currently there's a slow sequential step in binary relocation where all
strings of a binary are collected, with rpaths removed, and then
filtered for the old install root.
This is completely unnecessary, and also incorrect, since we replace
more than just the old install root in the prefix to prefix mapping. And
in fact the prefix to prefix mapping is parallel, and a single pass. So
even as an optimization, this filter makes no sense anymore.
Therefor we remove it
- single pass over the binary data matching all prefixes
- collect offsets and replacement strings
- do in-place updates with `fseek` / `fwrite`, since typically our
replacement touch O(few bytes) while the file is O(many megabytes)
- be nice: leave the file untouched if some string can't be
replaced
* Add patches for building clingo with MSVC
* Help python find clingo DLL
* If an executable is located in "C:\Program Files", Executable was
running into issues with the extra space. This quotes the exe
to ensure that it is treated as a single value.
Signed-off-by: Kiruya Momochi <65301509+KiruyaMomochi@users.noreply.github.com>
This commit extends the DSL that can be used in packages
to allow declaring that a package uses different build-systems
under different conditions.
It requires each spec to have a `build_system` single valued
variant. The variant can be used in many context to query, manipulate
or select the build system associated with a concrete spec.
The knowledge to build a package has been moved out of the
PackageBase hierarchy, into a new Builder hierarchy. Customization
of the default behavior for a given builder can be obtained by
coding a new derived builder in package.py.
The "run_after" and "run_before" decorators are now applied to
methods on the builder. They can also incorporate a "when="
argument to specify that a method is run only when certain
conditions apply.
For packages that do not define their own builder, forwarding logic
is added between the builder and package (methods not found in one
will be retrieved from the other); this PR is expected to be fully
backwards compatible with unmodified packages that use a single
build system.
Instead of looping over multiple regexes and the entire text file
contents, create a giant regex with all literal prefixes and do a single
pass over files to detect prefixes. Not only is a single pass faster,
it's also likely that the regex is compiled better, given that most
prefixes share a common ... prefix.
In the dfs code, flip edges so that `parent` means `from` and
`spec` means `to` in the direction of traversal. This makes it slightly
easier to write generic/composable code. For example when using visitors
where one visitor reverses direction, and another only cares about
accepting particular edges or not depending on whether the target node
is seen before, it would be good if the second visitor didn't have to
know whether the order was changed or not.
Use the same compression level as `gzip` (6) instead of what Python uses
(9).
The LLVM tarball takes 4m instead of 12m to create, and is <1% larger.
That's not worth the wait...
#32137 added an option to update() a BinaryCacheIndex with a
cooldown: repeated attempts within this cooldown would not
actually retry. However, the cooldown was not properly
tracked for failures (which is common when the mirror
does not store any binaries and therefore has no index.json).
This commit ensures that update(..., with_cooldown=True) will
also skip the update even if a failure has occurred within the
cooldown period.
Due to reuse concretization, we may generate DAGs with two occurrences
of the same package corresponding to distinct specs. This happens when
build dependencies are reused, since their dependencies are ignored in
concretization.
This caused a regression, for example: `spec['openssl']` would take the
'openssl' of the build dep `cmake`, instead of the direct `openssl`
dependency, simply because the edge to `cmake` was traversed first and
we do depth first traversal.
One solution that was discussed is to limit `spec[name]` to just direct
deps, or direct deps + transitive link deps, but this is too breaking.
Instead, this PR simply prioritizes transitive link and direct
build/run/test deps, and then falls back to a full DAG traversal. So,
it's just about order of iteration.
Scan the text files for relocatable prefixes *before* creating a tarball,
to reduce the amount of work to be done during install from binary
cache.
Co-authored-by: Harmen Stoppels <harmenstoppels@gmail.com>
Instead of showing
```
==> Error: Timed out waiting for a write lock.
```
show
```
==> Error: Timed out waiting for a write lock after 1.200ms and 4 attempts on file: /some/file
```
s.t. we actually get to see where acquiring a lock failed even when not
running in debug mode.
And use pretty time units everywhere, so we don't get 1.45e-9 seconds
but 1.450ns etc.
* backtraces without --debug
Currently `--debug` is too verbose and not-`--debug` gives to little
context about where exceptions are coming from.
So, instead, it'd be nice to have `spack --backtrace` and
`SPACK_BACKTRACE=1` as methods to get something inbetween: no verbose
debug messages, but always a full backtrace.
This is useful for CI, where we don't want to drown in debug messages
when installing deps, but we do want to get details where something goes
wrong if it goes wrong.
* completion
Currently `relocate_text` and `relocate_text_bin` are unsafe in the
sense that they run in parallel, and lead to races when modifying
different items pointing to the same inode.
This leads to the issue observed in #33453.
This PR:
1. Renames those functions to `unsafe_*` so people are aware
2. Adds logic to deal with hardlinks in current binary packages
3. Adds logic to deal with hardlinks when creating new binary tarballs,
so the install side doesn't have to de-dupe hardlinks.
4. Adds a test for 3
The assumption is that all our relocation logic preserves inodes. That
is, we should never copy a file, modify it, and then move it back. I
quickly verified, and its seems like this is true for (binary) text
relocation, as well as rpath patching in patchelf (even when the file
grows) and mach-o fixes.
* gitlab: Do not use root_spec['pkg_name'] anymore
For a long time it was fine to index a concrete root spec with the name
of a dependency in order to access the concrete dependency spec. Since
pipelines started using `--use-buildcache dependencies:only,package:never`
though, it has exposed a scheduling issue in how pipelines are
generated. If a concrete root spec depends on two different hashes of
`openssl` for example, indexing that root with just the package name
is ambiguous, so we should no longer depend on that approach when
scheduling jobs.
* env: make sure exactly one spec in env matches hash
When installing some/all specs from a buildcache, build edges are pruned
from those specs. This can result in a much smaller effective DAG. Until
now, `spack env depfile` would always generate a full DAG.
Ths PR adds the `spack env depfile --use-buildcache` flag that was
introduced for `spack install` before. This way, not only can we drop
build edges, but also we can automatically set the right buildcache
related flags on the specific specs that are gonna get installed.
This way we get parallel installs of binary deps without redundancy,
which is useful for Gitlab CI.
When downloading from binary cache not only replace RPATHs to dependencies, but
also text references to dependencies.
Example:
`autoconf@2.69` contains a text reference to the executable of its dependency
`perl`:
```
$ grep perl-5 /shared/spack/opt/spack/linux-amzn2-x86_64_v3/gcc-7.3.1/autoconf-2.69-q3lo/bin/autoreconf
eval 'case $# in 0) exec /shared/spack/opt/spack/linux-amzn2-x86_64_v3/gcc-7.3.1/perl-5.34.1-yphg/bin/perl -S "$0";; *) exec /shared/spack/opt/spack/linux-amzn2-x86_64_v3/gcc-7.3.1/perl-5.34.1-yphg/bin/perl -S "$0" "$@";; esac'
```
These references need to be replace or any package using `autoreconf` will fail
as it cannot find the installed `perl`.
Co-authored-by: Stephen Sachs <stesachs@amazon.com>
"spack install" will not update the binary index if given a concrete
spec, which causes it to fall back to direct fetches when a simple
index update would have helped. For S3 buckets in particular, this
significantly and needlessly slows down the install process.
This commit alters the logic so that the binary index is updated
whenever a by-hash lookup fails. The lookup is attempted again with
the updated index before falling back to direct fetches. To avoid
updating too frequently (potentially once for each spec being
installed), BinaryCacheIndex.update now includes a "cooldown"
option, and when this option is enabled it will not update more
than once in a cooldown window (set in config.yaml).
Co-authored-by: Tamara Dahlgren <35777542+tldahlgren@users.noreply.github.com>
* Add libpressio and dependencies; some of these packages are
maintained as forks of the original repositories and in those
cases the docstring mentions this.
* Add optional dependency in adios2 on libpressio
* cub package: set CUB_DIR environment variable for dependent
installations
* Clear R_HOME/R_ENVIRON before Spack installation (avoid sources
outside of Spack from affecting the installation in Spack)
* Rename dlib to dorian3d-dlib and update dependents; add new dlib
implementation. Pending an official policy on how to handle
packages with short names, reviewer unilaterally decided that
the rename was acceptable given that the new Spack dlib package
is referenced more widely (by orders of magnitude) than the
original
Co-authored-by: Samuel Li <shaomeng@users.noreply.github.com>
When installing an individual spec `spack --only=package --cache-only /xyz`
from a buildcache, Spack currently issues tons of warnings about
missing build deps (and their deps) in the database.
This PR disables these warnings, since it's fine to have a spec without
its build deps in the db (they are just "missing").
Currently `traverse_dependencies` fixes deptypes to traverse once and
for all in the recursion, but this is incorrect, since deptypes depend
on the node (e.g. if it's a dependency and cache-only, don't follow
build type edges, even if the parent is build from sources and needs
build deps.)
Support spackbot rebuilding all specs from source when asked (with "rebuild everything")
- Allow overriding --prune-dag cli opt with env var
- Use job variable to optionally prevent rebuild jobs early exit behavior
- ci rebuild: Use new install argument to insist deps are always installed from binary, but
package is only installed from source.
- gitlab: fix bug w/ untouched pruning
- ci rebuild: install from hash rather than json file
- When doing a "rebuild everything" pipeline, make sure that each install job only consumes
binary dependencies from the mirror being populated by the current pipeline. This avoids
using, e.g. binaries from develop, when rebuilding everything on a PR.
- When running a pipeline to rebuild everything, do not die because we generated a hash on
the broken specs list. Instead only warn in that case.
- bugfix: Replace broken no-args tty.die() with sys.exit(1)
Print a message of the form
```
Fetch mm:ss. Build: mm:ss. Total: mm:ss
```
when installing from buildcache.
Previously this only happened for source builds.
Currently "spack ci generate" chooses the first matching entry in
gitlab-ci:mappings to fill attributes for a generated build-job,
requiring that the entire configuration matrix is listed out
explicitly. This unfortunately causes significant problems in
environments with large configuration spaces, for example the
environment in #31598 (spack.yaml) supports 5 operating systems,
3 architectures and 130 packages with explicit size requirements,
resulting in 1300 lines of configuration YAML.
This patch adds a configuraiton option to the gitlab-ci schema called
"match_behavior"; when it is set to "merge", all matching entries
are applied in order to the final build-job, allowing a few entries
to cover an entire matrix of configurations.
The default for "match_behavior" is "first", which behaves as before
this commit (only the runner attributes of the first match are used).
In addition, match entries may now include a "remove-attributes"
configuration, which allows matches to remove tags that have been
aggregated by prior matches. This only makes sense to use with
"match_behavior:merge". You can combine "runner-attributes" with
"remove-attributes" to effectively override prior tags.
When a pipeline generation job is automatically failed because it
generated jobs for specs known to be broken on develop, print better
information about the broken specs that were encountered. Include
at a minimum the hash and the url of the job whose failure caused it
to be put on the broken specs list in the first place.
* env depfile: allow deps only install
- Refactor `spack env depfile` to use a Jinja template, making it a bit
easier to follow as a human being.
- Add a layer of indirection in the generated Makefile through an
`<prefix>/.install-deps/<hash>` target, which allows one to specify
different options when installing dependencies. For example, only
verbose/debug mode on when installing some particular spec:
```
$ spack -e my_env env depfile -o Makefile --make-target-prefix example
$ make example/.install-deps/<hash> -j16
$ make example/.install/<hash> SPACK="spack -d" SPACK_INSTALL_FLAGS=--verbose -j16
```
This could be used to speed up `spack ci rebuild`:
- Parallel install of dependencies from buildcache
- Better readability of logs, e.g. reducing verbosity when installing
dependencies, and splitting logs into deps.log and current_spec.log
* Silence please!
* spack.compiler.Compiler: introduce prefix property
We currently don't really have something that gives the GCC install
path, which is used by many LLVM-based compilers (llvm, llvm-amdgpu,
nvhpc, ...) to fix the GCC toolchain once and for all.
This `prefix` property is dynamic in the sense that it queries the
compiler itself. This is necessary because it's not easy to deduce the
install path from the `cc` property (might be a symlink, might be a
filename like `gcc` which works by having the compiler load a module
that sets the PATH variable, might be a generic compiler wrapper based
on environment variables like on cray...).
With this property introduced, we can clean up some recipes that have
the logic repeated for GCC.
* intel-oneapi-compilers: set --gcc-sysroot to %gcc prefix
Caches used by repositories don't reference the global spack.repo.path instance
anymore, but get the repository they refer to during initialization.
Spec.virtual now use the index, and computation done to compute the index
use Repository.is_virtual_safe.
Code to construct mock packages and mock repository has been factored into
a unique MockRepositoryBuilder that is used throughout the codebase.
Add debug print for pushing and popping config scopes.
Changed spack.repo.use_repositories so that it can override or not previous repos
spack.repo.use_repositories updates spack.config.config according to the modifications done
Removed a peculiar behavior from spack.config.Configuration where push would always
bubble-up a scope named command_line if it existed
Resolves#31782
With this change, if a spec is concrete after parsing (e.g. spec.yaml
or /hash-based), then it is not disambiguated (a process which requires
(a) that the spec be installed and (b) that it be part of the
currently-active environment).
This commit allows you to:
* Diff specs from an environment regardless of whether they have
been installed (more useful for projection/matrix-based envs)
* Diff specs read from .yaml files which may or may not be entirely
different installations of Spack
Co-authored-by: Massimiliano Culpo <massimiliano.culpo@gmail.com>
* Docs: Getting Started Dependencies
Finally document what one needs to install to use Spack on
Linux and Mac :-)
With <3 for minimal container users and my colleagues with
their fancy Macs.
* Debian Update Packages: GCC, Python
- build-essential: includes gcc, g++ (thx Cory)
- Python: add python3-venv, python3-distutils (thx Pradyun)
* Add RHEL8 Dependencies
* filter_file: introduce argument 'start_at'
* autotools: extend patching of the libtool script
* autotools: refactor _patch_usr_bin_file
* autotools: improve readability of the filtering
* autotools: keep the modification time of the configure scripts
* autotools: do not try to patch directories
* autotools: explain libtool patching for posterity
Co-authored-by: Massimiliano Culpo <massimiliano.culpo@gmail.com>
Co-authored-by: Harmen Stoppels <harmenstoppels@gmail.com>
Remove `module-info mode load` condition that prevents auto-unloading when autoloading is enabled. It looks like this condition was added to work around an issue in environment-modules that is no longer necessary.
Add quotes to make is-loaded happy
When concurrent misc_cache provider index rebuilds happen, try to
rebuild it only once, so we don't exceed misc_cache lock timeout.
For example, when using `spack env depfile`, with no previous
misc_cache, running `make -f depfile -j8` could run at most 8 concurrent
`spack install` locking on misc_cache to rebuild the provider index. If
one rebuild takes 30s, before this fix, the "worst" lock could wait up
to 30s * 7, easily exceeding misc_cache lock timeout. Now, the "worst"
lock would take 30s * 1 + ~1s * 6.
Currently, module changes from `setup_dependent_package` are applied only to the module of the package class, but not to any parent classes' modules between the package class module and `spack.package_base`.
In this PR, we create a custom class to accumulate module changes, and apply those changes to each class that requires it. This design allows us to code for a single module, while applying the changes to multiple modules as needed under the hood, without requiring the user to reason about package inheritance.
* find/list: display package counts last
We have over 6,600 packages now, and `spack list` still displays the number of packages
before it lists them all. This is useless for large sets of results (e.g., with no args)
as the number has scrolled way off the screen before you can see it. The same is true
for `spack find` with large installations.
This PR changes `spack find` and `spack list` so that they display the package count
last.
* add some quick testing
Co-authored-by: Danny McClanahan <1305167+cosmicexplorer@users.noreply.github.com>
Allow environment variables and spack-specific path substitution variables (e.g. `$spack`) to be
used in the paths associated with develop specs, while maintaining the ability to keep those
paths relative to the environment rather than the working directory.
Install: Add use-buildcache option to install
* Allow differentiating between top level packages and dependencies when
determining whether to install from the cache or not.
* Add unit test for --use-buildcache
* Use metavar to display use-buildcache options.
* Update spack-completion
Make it possible to install the Clingo package on Windows; this
also provides a means to use Clingo with Spack on Windows.
This includes
* A new "winbison" package: Windows has a port of bison and flex where
the two packages are grouped together. Clingo dependencies have been
updated to use winbison on Windows and bison elsewhere (this avoids
complicating the existin bison/flex packages until we can add support
for implied virtuals).
* The CMake build system was incorrectly converting CMAKE_INSTALL_PREFIX
to POSIX format.
* The re2c package has been modified to use CMake on Windows; for now
this is done by overloading the configure/build/install methods to
perform CMake-appropriate operations; the package should be refactored
once support for multiple build systems in one Package is available.
This commit fixes#27027.
The root cause of the issue is that the `SPACK_OLD_PROMPT` variable
was evaluated in string interpolation regardless of whether the
guard condition above evaluates to true or false. This commit uses
the `eval` keyword to defer evaluation until the command is executed.
Co-authored-by: Alexander Hornburg <alexande@xilinx.com>
Spack currently depends on parsing filenames of downloaded files to
determine what type of archive they are and how to decompress them.
This commit adds a preliminary check based on magic numbers to
determine archive type (but falls back on name parsing if the
extension type cannot be determined).
As part of this work, this commit also enables decompression of
.tar.xz-compressed archives on Windows.
Include exception info related to url retrieval in debug messages
which otherwise would be swallowed. This is intended to be useful
for detecting if CA configuration interferes with downloads from
HTTPS links.
