f_bfree space might not be usable. In practice we see failures in big
disks because f_bfree is over 2GB and f_bavail is 0. We get ENOSPC if
writing on disk then.

We plan to make cache incomplete. That is some blobs are missing.  For
most of cases we will delete references when requested if they are
incomplete. But there will be corner cases where objects are removed
after the reference is requested.

The file is already a temporary file and does not need copy.  ENOSPC
is thrown during that copy in issue #609.

Fixes #678.

This locks the temporary object file so that cleanup does not need to
be done for every write.

This also remove references when some objects are missing. This is in
preparation for the move from reference to object garbage collection.

When there is less than 2GB left, it cleans up have 10GB available.
These values are configurable.

Cleaning up in parallel might slow down the cleaning process

Fix cleanup of cache in server when disk is full

Closes #678

See merge request !830

push_message and push_directory will both raise assertion failures if
they fail to send the digest to the server. Checking the digest
returned by each only tests that the content was hashed locally, which
cannot reasonably fail.

_sandboxremote.py: Remove unnecessary tests.

Closes #786

See merge request !976

Fixes #723.

Add support for .netrc in remote/tar/zip plugins

Closes #723

See merge request !908

Angelos Evripiotis authored Nov 29, 2018 and Jürg Billeter committed Nov 30, 2018

It seems that the regenerated man pages haven't been committed for some
time. Add them all in one lump now.

Had to manually strip the version number from the generated files.

Had to manually fix the NAME field of bst.1 to not be 'bst - None'.

Oddly had to invoke click-man directly in order to get the changes to
be applied, the usual setup.py route mentioned in contributing.rst
didn't work.

Eyeballed the result of invoking man on each of the commands, seems
good.

man/: update with changes since Apr 2018

See merge request !981

Tom Pollard authored Nov 28, 2018 and Jürg Billeter committed Nov 30, 2018

Adds helper function _buildtree_pull_required() to determine if a
pullqueue should be constructed, for commands outside of bst pull
and build where it is determined that an element's buildtree
artifact is to be required given the respective semantics and
config. Utilised in push() to attempt to mitigate skipping the push
of partial elements without the user having to have preceded it with
an explicit pull.

cli.py: Add new behaviour to push command description

element.py: Move _cached_buildtree() to be non local private method,
use _KeyStrength types to reduce duplication.

tests/integration/pullbuildtrees.py also updated to cover this
use-case.

_stream.py: Ability to pull missing buildtrees outside of pull/build

Closes #774

See merge request !978

_loader/loader.py: Added Suffix check and warning for named elements and add helper function

buildcheckout.py: Add mandatory .bst suffix tests
completions.py: Add test for fail on invalid suffix
Added required files for testing
integration/source-determinism.py: renamed test elements to end with .bst
loader/__init__.py: initialised context properly with load and a message handler

Mandatory .bst suffix

See merge request !967

This was previously missing, causing our CI to forget to rebuild the sessions
when updating the documentation.

This fixes issue #778

MANIFEST.in: Include the sources for rebuilding the documentation session HTML

Closes #778

See merge request !984

Chandan Singh authored Sep 26, 2018 and Benjamin Schubert committed Dec 03, 2018

Currently, `bst checkout --deps none` command always produces empty
output. Fix this issue and add regression test for the same.

- element_enums.py: Add Scope.NONE.

- element.py: Ensure Scope.NONE works correctly in addition to
  Scope.RUN/Scope.ALL in Element.dependencies() and Element.search().

- tests/frontend/buildcheckout.py: Fix tests for `--deps none`.

Fixes #670.

Chandan Singh authored Sep 21, 2018 and Benjamin Schubert committed Dec 03, 2018

As discussed in
https://mail.gnome.org/archives/buildstream-list/2018-September/msg00064.html,
add `--deps build` option to `bst checkout`. This will allow users to
checkout the all build dependencies of a given element using a single
command.

- _frontend/cli.py: Add `--deps build` option for `bst checkout`.

- element.py: Support `deps='build'` in Element._prepare_sandbox().

- tests/frontend/buildcheckout.py: Ensure `--deps build` works as
  expected.

Before we would have a intricate logics with multiple arguments
that might get ignored.

This simplifies the design and introduces a bool `shell` instead
of having two different variables concerned about scope

Add `--deps build` option to `bst checkout`

Closes #670

See merge request !819

Javier Jardón authored Dec 02, 2018 and Tristan Van Berkom committed Dec 03, 2018

See #783

git: warn about checkout submodules  by default

See merge request !983

This is required because if there are projects with the same name
loaded, that have elements in the same path, with the same contents,
they are considered the same, despite actually belonging to different
Projects.

This fixes buildstream issue #795

James Ennis authored Nov 30, 2018 and Jonathan Maw committed Dec 03, 2018

Test required to close #795 - We should be able to build a project which refers
to the same junction twice, but with two different filenames.

_yamlcache.py: Use a project's junction name if present

Closes #795

See merge request !980

Commit 3fa79d8d, part of an initiative for caching of the failed builds,
introduced a call to Element._update_state() after a job completes and before
entering the error handling harness intended for handling plugin raised errors.

Element._update_state() can result in triggering plugin code to
run, so this is incorrect, and causes raised errors to crash BuildStream
if they happen here.

After analyzing the code, it appears that this additional call to
Element._update_state() is unneeded, and was only added because the
state needs to be updated for a failure as well as a success.

Instead, we now have the BuildQueue call Element._assemble_done()
unconditionally, regardless of whether the build was successful
or not, which has the same effect and also reads better.

In addition, added a FIXME comment that we are still conditionally
updating the artifact cache size from BuildQueue.done() only
if the build is successful, which is incorrect because failed builds
also increase the local artifact cache size - to fix this we need
to communicate the added artifact size through Element._assemble()
regardless of whether the build succeeded or failed.

_scheduler/queues/queue.py: Don't call update state outside of error handling harness

See merge request !988

Instead of tag information being fetched which can change with time,
they are tracked and saved in the projects.refs/.bst. Then we re-tag
automatically the closest tag so that `git describe` works and is
reproducible.

This new feature is opt-in with the new `track-tags` configuration,
and must be used to fix modules which are broken by our new policy
of omitting the `.git/` repository when staging git sources.

This fixes issue #487

Track of git tags and save them to reproduce minimum shallow repository

Closes #487

See merge request !906

In element, platform asks for host os and architecture to default to when
SandboxConfig is initialised.
This changes element cache keys so those have been updated in the tests.

Also update tests to be consistent with this

These are contained within the platform field.

Same format as OptionArch tests, but with OS's.

  o _platform/linux.py: Add linux32 flag to send to sandbox bwrap when the
    build arch is x86-32 and the machines arch is x86-64 or similarly with
    aarch32 and aarch64.
  o sandbox/_sandboxbwrap.py: Use flag passed to start bwrap command with
    linux32 if set.

Execution environment reqs

Closes #775

See merge request !969

Get cascache.py to use this

Stops unneccesary pushing of builds that have already been built, just checks
the action cache to begin with.

Fixes #628

The endpoint may not necessarily be the same as execution-service endpoint.

Part of #628

Part of #628

Remote-execution client flow optimisation

Closes #628

See merge request !982

Instead of duplicating the whole test body, run it twice while
checking for a warning or an error depending on the parameter.

This also ensures it is exposed via the main buildstream __init__.py
file, fixes the imports of CoreWarnings, and adjusts the documentation
links to point to the right place.

Now the commands have documentation linkability so we can link to
them when referring to them anywhere in the docs. This reference
was previously referring to a bogus non-existent `build-stream track`,
maybe this dates way back to the days before we named the frontend `bst`.

Fixes the configurable warning part of this to properly link
to the project.conf documentation describing what configurable
warnings actually are.

Also fix the plugin raise the actual configurable warning which
it advertizes that it raises; which is 'inconsistent-submodule',
not 'inconsistent-submodules'.

Change 'INCONSISTENT_SUBMODULE' for 'WARN_INCONSISTENT_SUBMODULE' to
make things a bit more readable and comprehensive.

Misc cleanups

See merge request !993