Variant of this kind don't have a list of possible
values encoded in the ASP facts. Since all we have
is a validator the list of possible values just includes
just the default value and possibly the value passed
from packages.yaml or cli.
This is done after the builder has actually built
the specs, to respect the semantics use with the
old concretizer.
Later we could move this to the solver as
a multivalued variant.
This is done after the builder has actually built
the specs, to respect the semantics use with the
old concretizer.
A better approach is to substitute the spec
directly in concretization.
The "none" variant value cannot be combined with
other values.
The '*' wildcard matches anything, including "none".
It's thus relevant in queries, but disregarded in
concretization.
- The test on concretization of anonymous dependencies
has been fixed by raising the expected exception.
- The test on compiler bootstrap has been fixed by
updating the version of GCC used in the test.
Since gcc@2.0 does not support targets later than
x86_64, the new concretizer was looking for a
non-existing spec, i.e. it was correctly trying
to retrieve 'gcc target=x86_64' instead of
'gcc target=core2'.
- The test on gitlab CI needed an update of the target
This commit adds support for specifying rules in
packages.yaml that refer to virtual packages.
The approach is to normalize in memory each
configuration and turn it into an equivalent
configuration without rules on virtual. This
is possible if the set of packages to be handled
is considered fixed.
The weight of the target used in concretization is, in order:
1. A specific per package weight, if set in packages.yaml
2. Inherited from the parent, if possible
3. The default target weight (always set)
Generate facts on externals by inspecting
packages.yaml. Added rules in concretize.lp
Added extra logic so that external specs
disregard any conflict encoded in the
package.
In ASP this would be a simple addition to
an integrity constraint:
:- c1, c2, c3, not external(pkg)
Using the the Backend API from Python it
requires some scaffolding to obtain a default
negated statement.
Conflict rules from packages are added as integrity
constraints in the ASP formulation. Most of the code
to generate them has been reused from PyclingoDriver.rules
The new concretizer and the old concretizer solve constraints
in a different way. Here we ensure that a SpackError is raised,
instead of a specific error that made sense in the old concretizer
but probably not in the new.
Instead of python callbacks, use cardinality constraints for package
versions. This is slightly faster and has the advantage that it can be
written to an ASP program to be executed *outside* of Spack. We can use
this in the future to unify the pyclingo driver and the clingo text
driver.
This makes use of add_weight_rule() to implement cardinality constraints.
add_weight_rule() only has a lower bound parameter, but you can implement
a strict "exactly one of" constraint using it. In particular, wee want to
define:
1 {v1; v2; v3; ...} 1 :- version_satisfies(pkg, constraint).
version_satisfies(pkg, constraint) :- 1 {v1; v2; v3; ...} 1.
And we do that like this, for every version constraint:
atleast1(pkg, constr) :- 1 {version(pkg, v1); version(pkg, v2); ...}.
morethan1(pkg, constr) :- 2 {version(pkg, v1); version(pkg, v2); ...}.
version_satisfies(pkg, constr) :- atleast1, not morethan1(pkg, constr).
:- version_satisfies(pkg, constr), morethan1.
:- version_satisfies(pkg, constr), not atleast1.
v1, v2, v3, etc. are computed on the Python side by comparing every
possible package version with the constraint.
Computing things like this has the added advantage that if v1, v2, v3,
etc. comprise *all* possible versions of a package, we can just omit the
rules for the constraint under consideration. This happens pretty
frequently in the Spack mainline.
- [x] Solver now uses the Python interface to clingo
- [x] can extract unsatisfiable cores from problems when things go wrong
- [x] use Python callbacks for versions instead of choice rules (this may
ultimately hurt performance)
There are now three parts:
- `SpackSolverSetup`
- Spack-specific logic for generating constraints. Calls methods on
`AspTextGenerator` to set up the solver with a Spack problem. This
shouln't change much from solver backend to solver backend.
- ClingoDriver
- The solver driver provides methods for SolverSetup to generates an ASP
program, send it to `clingo` (run as an external tool), and parse the
output into function tuples suitable for `SpecBuilder`.
- The interface is generic and should not have to change much for a
driver for, say, the Clingo Python interface.
- SpecBuilder
- Builds Spack specs from function tuples parsed by the solver driver.
The original implementation was difficult to read, as it only had
single-letter variable names. This converts all of them to descriptive
names, e.g., P -> Package, V -> Virtual/Version/Variant, etc.
To handle unknown compilers propely in tests (and elsewhere), we need to
add unknown compilers from the spec to the list of possible compilers.
Rework how the compiler list is generated and includes compilers from
specs if the existence check is disabled.
Specs like hdf5 ^mpi were unsatisfiable because we added a requierment
for `node("mpi").`. This can't be resolved because "mpi" is not a
package.
- [x] Introduce `virtual_node()`, which says *some* provider must be in
the DAG.
This adds compiler flags to the ASP solve so that we can have conditions
based on them in the solve. But, it keeps order out of the solve to
avoid unneeded complexity and combinatorial explosions.
The solver determines which flags are on a spec, but the order is
determined by DAG precedence (childrens' flags take precedence over
parents' and are added on the right) and order (order flags were
specified on the command line is respected).
The solver is responsible for determining when to propagate flags, when
to inheit them from other nodes, when to take them from compiler
preferences, etc.
Weight microarchitectures and prefers more rercent ones. Also disallow
nodes where the compiler does not support the selected target.
We should revisit this at some point as it seems like if I play around
with the compiler support for different architectures, the solver runs
very slowly. See notes in comments -- the bad case was gcc supporting
broadwell and skylake with clang maxing out at haswell.
We didn't have a cardinality constraint for multi-valued variants, so the
solver wasn't filling them in.
- [x] add a requirement for at least one value for multi-valued variants
Variants like `cpu_target` on `openblas` don't have defineed values, but
they have a default. Ensure that the default is always a possible value
for the solver.
Spack was generating the same dependency connstraints twice in the output ASP:
```
declared_dependency("abinit", "hdf5", "link")
:- node("abinit"),
variant_value("abinit", "mpi", "True"),
variant_value("abinit", "mpi", "True").
```
This was because `AspFunction` was modifying itself when called.
- [x] fix `AspFunction` so that every call returns a new object
- [x] Add support for packages.yaml and command-line compiler preferences.
- [x] Rework compiler version propagation to use optimization rather than
hard logic constraints
Technically the ASP output order does not matter, but it's hard to diff
two different solve fomulations unless we order it.
- [x] make sure ASP output is emitted in a deterministic order (by
sorting all hash keys)
This needs more thought, as I am pretty sure the weights are not correct.
Or, at least, I'm not convinced that they do what we want in all cases.
See note in concretize.lp.
Solver now prefers newer versions like the old concretizer. Prefer
package preferences from packages.yaml, preferred=True, package
definition, and finally each version itself.
Competition output only prints out one model, so we do not have to
unnecessarily parse all the non-optimal models. We'll just look at the
best model and bring that in.
In practice, this saves a lot of JSON parsing and spec construction time.
Clingo actually has an option to output JSON -- use that instead of
parsing the raw otuput ourselves.
This also allows us to pick the best answer -- modify the parser to
*only* construct a spec for that one rather than building all of them
like we did before.
- Instead of using default logic, handle variant defaults by minimizing
the number of non-default variants in the solution.
- This actually seems to be pretty fast, and it fixes the long-standing
issue that writing this:
spack install hdf5 ^mpich
will fail if you don't specify hdf5+mpi. With optimization and
allowing enums to be enumerated, the solver seems to be able to quickly
discover that +mpi is the only way hdf5 can depend on mpich, and it
forces the switch to be thrown.
Use '1 { version(x); version(y); version(z) } 1.' instead of declaring
conflicts for non-matching versions. This keeps the sense of version
clauses positive, which will allow them to be used more easily in
conditionals later.
Also refactor `spec_clauses()` method to return clauses that can be used
in conditions, etc. instead of just printing out facts.
- This handles setting the compiler and falling back to a default
compiler, as well as providing default values for compilers/compiler
versions.
- Versions still aren't quite right -- you can't properly override
versions on compiler specs.
- Model architecture default settings and propagation off of variants
- Leverage ASP default logic to set architecture to default if it's not
set otherwise.
- Move logic out of Python and into concretize.lp as first-order rules.
We are relying on default logic in the variant handling in that we set a
default value if we never see `variant_set(P, V, X)`.
- Move the logic for this into `concretize.lp` instead of generating it
for every package.
- For programs that don't have explicit variant settings, clingo warns
that variant_set(P, V, X) doesn't appear in any rule head, because a
setting is never generated.
- Specifically suppress this warning.
- moving the dump logic into spack.solver.asp.solve() allows us to print
out useful debug info sooner
- prior approach required a successful solve to print out anyhting.
According to the documentation for spack and pkg-config,
$view/share/pkgconfig should also be a valid place to look
for package config files. This commit ensures that when
spack activate env $dir is called, the environment has this
directory in PKG_CONFIG_PATH.
As of #13100, Spack installs the dependencies of a _single_ spec in parallel.
Environments, when installed, can only get parallelism from each individual
spec, as they're installed in order. This PR makes entire environments build
in parallel by extending Spack's package installer to accept multiple root
specs. The install command and Environment class have been updated to use
the new parallel install method.
The specs and kwargs for each *uninstalled* package (when not force-replacing
installations) of an environment are collected, passed to the `PackageInstaller`,
and processed using a single build queue.
This introduces a `BuildRequest` class to track install arguments, and it
significantly cleans up the code used to track package ids during installation.
Package ids in the build queue are now just DAG hashes as you would expect,
Other tasks:
- [x] Finish updating the unit tests based on `PackageInstaller`'s use of
`BuildRequest` and the associated changes
- [x] Change `environment.py`'s `install_all` to use the `PackageInstaller` directly
- [x] Change the `install` command to leverage the new installation process for multiple specs
- [x] Change install output messages for external packages, e.g.:
`[+] /usr` -> `[+] /usr (external bzip2-1.0.8-<dag-hash>`
- [x] Fix incomplete environment install's view setup/update and not confirming all
packages are installed (?)
- [x] Ensure externally installed package dependencies are properly accounted for in
remaining build tasks
- [x] Add tests for coverage (if insufficient and can identity the appropriate, uncovered non-comment lines)
- [x] Add documentation
- [x] Resolve multi-compiler environment install issues
- [x] Fix issue with environment installation reporting (restore CDash/JUnit reports)
This change makes improvements to the `spack ci rebuild` command
which supports running gitlab pipelines on PRs from forks. Much
of this has to do with making sure we can run without the secrets
previously required for running gitlab pipelines (e.g signing key,
aws credentials, etc). Specific improvements in this PR:
Check if spack has precisely one signing key, and use that information
as an additional constraint on whether or not we should attempt to sign
the binary package we create.
Also, if spack does not have at least one public key, add the install
option "--no-check-signature"
If we are running a pipeline without any profile or environment
variables allowing us to push to S3, the pipeline could still
successfully create a buildcache in the artifacts and move on. So
just print a message and move on if pushing either the buildcache
entry or cdash id file to the remote mirror fails.
When we attempt to generate a pacakge or gpg key index on an S3
mirror, and there is nothing to index, just print a warning and
exit gracefully rather than throw an exception.
Support the use of PR-specific mirrors for temporary binary pkg
storage. This will allow quality-of-life improvement for developers,
providing a place to store binaries over the lifetime of a PR, so
that they must only wait for packages to rebuild from source when
they push a new commit that causes it to be necessary.
Replace two-pass install with a single pass and the new option:
--require-full-hash-match. Doing this also removes the need to
save a copy of the spack.yaml to be copied over the one spack
rewrites in between the two spack install passes.
Work around a mirror configuration issue caused by using
spack.util.executable to do the package installation.
* Update pipeline trigger jobs for PRs from forks
Moving to PRs from forks relies on external synchronization script
pushing special branch names. Also secrets will only live on the
spack mirror project, and must be propagated to the E4S project via
variables on the trigger jobs.
When this change is merged, pipelines will not run until we update
the "Custom CI configuration path" in the Gitlab CI Settings, as the
name of the file has changed to better reflect its purpose.
* Arg to MirrorCollection is used exclusively, so add main remote mirror to it
* Compute full hash less frequently
* Add tests covering index generation error handling code
Since #11598 sbang has been installed within the install_tree. This doesn’t play
nicely with install_tree padding, since sbang can’t do its job if it is installed in a
long path (this is the whole point of sbang).
This PR changes the padding specification. Instead of $padding inside paths,
we now have a separate `padding:` field in the `install_tree` configuration.
Previously, the `install_tree` looked like this:
```
/path/to/opt/spack_padding_padding_padding_padding_padding/
bin/
sbang
.spack-db/
...
linux-rhel7-x86_64/
...
```
```
This PR updates things to look like this:
/path/to/opt/
bin/
sbang
spack_padding_padding_padding_padding_padding/
.spack-db/
...
linux-rhel7-x86_64/
...
So padding is added at the start of all install prefixes *within* the unpadded
root. The database and all installations still go under the padded root.
This ensures that `sbang` is in the shorted possible path while also allowing
us to make long paths for relocatable binaries.
As of #18205, all packages must be pickle-able to be installed by
Spack.
This adds a test to check that each package can be pickled. If any
package fails to pickle, the test keeps going and collects the names
of all failed packages; it then takes the first one that failed and
attempts to re-pickle it, generating the full stack trace for the
failed pickle attempt.
Spack creates a separate process to do package installation. Different
operating systems and Python versions use different methods to create
it but up until Python 3.8 both Linux and Mac OS used "fork" (which
duplicates process memory, file descriptor table, etc.).
Python >= 3.8 on Mac OS prefers creating an entirely new process
(referred to as the "spawn" start method) because "fork" was found to
cause issues (in other words "spawn" is the default start method used
by multiprocessing.Process). Spack was dependent on the particular
behavior of fork to replicate process memory and transmit file
descriptors.
This PR refactors the Spack internals to support starting a child
process with the "spawn" method. To achieve this, it makes the
following changes:
- ensure that the package repository and other global state are
transmitted to the child process
- ensure that file descriptors are transmitted to the child process in
a way that works with multiprocessing and spawn
- make all the state needed for the build process and tests picklable
(package, stage, etc.)
- move a number of locally-defined functions into global scope so that
they can be pickled
- rework tests where needed to avoid using local functions
This PR also reworks sbang tests to work on macOS, where temporary
directories are deeper than the Linux sbang limit. We make the limit
platform-dependent (macOS supports 512-character shebangs)
See: #14102
In compiler bootstrapping pipelines, we add an artificial dependency
between jobs for packages to be built with a bootstrapped compiler
and the job building the compiler. To find the right bootstrapped
compiler for each spec, we compared not only the compiler spec to
that required by the package spec, but also the architectures of
the compiler and package spec.
But this prevented us from finding the bootstrapped compiler for a
spec in cases where the architecture of the compiler wasn't exactly
the same as the spec. For example, a gcc@4.8.5 might have
bootstrapped a compiler with haswell as the architecture, while the
spec had broadwell. By comparing the families instead of the architecture
itself, we know that we can build the zlib for broadwell with the gcc for
haswell.
Currently, full JSON output is the only machine readable option for `spack find`
in an environment.
`spack find --format` is also designed to be machine readable, but we print extra
headers in environments.
-[x] don't print headers in `spack find` output when in an environment
