def test_partition(self):
a = MockTarget('a', [], 1)
b = MockTarget('b', [a], 1)
c = MockTarget('c', [b], 1)
d = MockTarget('d', [c, a], 1)
e = MockTarget('e', [d], 1)
targets = [a, b, c, d, e]
all_vts = self.cache_manager._sort_and_validate_targets(targets)
[print_vt(vt) for vt in all_vts]
print ''
invalid_vts = filter(lambda vt: not vt.valid, all_vts)
self.assertEquals(5, len(invalid_vts))
self.assertEquals(5, len(all_vts))
vts_targets = [vt.targets[0] for vt in all_vts]
self.assertEquals(set(targets), set(vts_targets))
ic = InvalidationCheck(all_vts, [], 3)
partitioned = ic.all_vts_partitioned
[print_vt(vt) for vt in partitioned]
# NOTE(ryan): several correct partitionings are possible, but in all cases 4 1-source targets will be added to the
# first partition before it exceeds the limit of 3, and the final target will be in a partition by itself.
self.assertEquals(2, len(partitioned))
self.assertEquals(4, len(partitioned[0].targets))
self.assertEquals(1, len(partitioned[1].targets))
def invalidated(self, targets, only_buildfiles=False, invalidate_dependents=False,
partition_size_hint=sys.maxint):
"""Checks targets for invalidation, first checking the artifact cache.
Subclasses call this to figure out what to work on.
targets: The targets to check for changes.
only_buildfiles: If True, then only the target's BUILD files are checked for changes, not
its sources.
invalidate_dependents: If True then any targets depending on changed targets are invalidated.
partition_size_hint: Each VersionedTargetSet in the yielded list will represent targets
containing roughly this number of source files, if possible. Set to
sys.maxint for a single VersionedTargetSet. Set to 0 for one
VersionedTargetSet per target. It is up to the caller to do the right
thing with whatever partitioning it asks for.
Yields an InvalidationCheck object reflecting the (partitioned) targets.
If no exceptions are thrown by work in the block, the build cache is updated for the targets.
Note: the artifact cache is not updated. That must be done manually.
"""
with self.context.new_workunit('invalidation'):
extra_data = []
extra_data.append(self.invalidate_for())
for f in self.invalidate_for_files():
extra_data.append(hash_file(f))
cache_manager = CacheManager(self._cache_key_generator,
self._build_invalidator_dir,
invalidate_dependents,
extra_data,
only_externaldeps=only_buildfiles)
invalidation_check = cache_manager.check(targets, partition_size_hint)
# See if we have entire partitions cached.
if invalidation_check.invalid_vts and self._artifact_cache and \
self.context.options.read_from_artifact_cache:
with self.context.new_workunit('cache'):
all_cached_targets = []
partitions_to_check = \
[vt for vt in invalidation_check.all_vts_partitioned if not vt.valid]
cached_partitions, uncached_partitions = self.check_artifact_cache(partitions_to_check)
for vt in cached_partitions:
for t in vt.targets:
all_cached_targets.append(t)
# See if we have any individual targets from the uncached partitions.
vts_to_check = [vt for vt in itertools.chain.from_iterable(
[x.versioned_targets for x in uncached_partitions]) if not vt.valid]
cached_targets, uncached_targets = self.check_artifact_cache(vts_to_check)
for vt in cached_targets:
all_cached_targets.append(vt.target)
if all_cached_targets:
# Do some reporting.
for t in all_cached_targets:
self.context.run_tracker.artifact_cache_stats.add_hit('default', t)
self._report_targets('Using cached artifacts for ', all_cached_targets, '.')
# Now that we've checked the cache, re-partition whatever is still invalid.
if uncached_targets:
for vts in uncached_targets:
self.context.run_tracker.artifact_cache_stats.add_miss('default', vts.target)
self._report_targets('No cached artifacts for ',
[vt.target for vt in uncached_targets], '.')
invalidation_check = \
InvalidationCheck(invalidation_check.all_vts, uncached_targets, partition_size_hint)
# Do some reporting.
targets = []
sources = []
num_invalid_partitions = len(invalidation_check.invalid_vts_partitioned)
for vt in invalidation_check.invalid_vts_partitioned:
targets.extend(vt.targets)
sources.extend(vt.cache_key.sources)
if len(targets):
msg_elements = ['Invalidated ',
items_to_report_element([t.address.reference() for t in targets], 'target')]
if len(sources) > 0:
msg_elements.append(' containing ')
msg_elements.append(items_to_report_element(sources, 'source file'))
if num_invalid_partitions > 1:
msg_elements.append(' in %d target partitions' % num_invalid_partitions)
msg_elements.append('.')
self.context.log.info(*msg_elements)
# Yield the result, and then mark the targets as up to date.
yield invalidation_check
if not self.dry_run:
for vt in invalidation_check.invalid_vts:
vt.update() # In case the caller doesn't update.
def invalidated(self,
targets,
only_buildfiles=False,
invalidate_dependents=False,
partition_size_hint=sys.maxint):
"""Checks targets for invalidation, first checking the artifact cache.
Subclasses call this to figure out what to work on.
targets: The targets to check for changes.
only_buildfiles: If True, then only the target's BUILD files are checked for changes,
not its sources.
invalidate_dependents: If True then any targets depending on changed targets are invalidated.
partition_size_hint: Each VersionedTargetSet in the yielded list will represent targets
containing roughly this number of source files, if possible. Set to
sys.maxint for a single VersionedTargetSet. Set to 0 for one
VersionedTargetSet per target. It is up to the caller to do the right
thing with whatever partitioning it asks for.
Yields an InvalidationCheck object reflecting the (partitioned) targets.
If no exceptions are thrown by work in the block, the build cache is updated for the targets.
Note: the artifact cache is not updated, that must be done manually.
"""
extra_data = []
extra_data.append(self.invalidate_for())
for f in sorted(self.invalidate_for_files()):
sha = hashlib.sha1()
with open(f, "rb") as fd:
sha.update(fd.read())
extra_data.append(sha.hexdigest())
cache_manager = CacheManager(self._cache_key_generator,
self._build_invalidator_dir,
invalidate_dependents,
extra_data,
only_externaldeps=only_buildfiles)
initial_invalidation_check = cache_manager.check(
targets, partition_size_hint)
# See if we have entire partitions cached.
partitions_to_check = \
[vt for vt in initial_invalidation_check.all_vts_partitioned if not vt.valid]
cached_partitions, uncached_partitions = self.check_artifact_cache(
partitions_to_check)
# See if we have any individual targets from the uncached partitions.
vts_to_check = [
vt for vt in itertools.chain.from_iterable(
[x.versioned_targets for x in uncached_partitions])
if not vt.valid
]
cached_targets, uncached_targets = self.check_artifact_cache(
vts_to_check)
# Now that we've checked the cache, re-partition whatever is still invalid.
invalidation_check = \
InvalidationCheck(initial_invalidation_check.all_vts, uncached_targets, partition_size_hint)
# Do some reporting.
num_invalid_partitions = len(
invalidation_check.invalid_vts_partitioned)
num_invalid_targets = 0
num_invalid_sources = 0
for vt in invalidation_check.invalid_vts:
if not vt.valid:
num_invalid_targets += len(vt.targets)
num_invalid_sources += vt.cache_key.num_sources
if num_invalid_partitions > 0:
self.context.log.info('Operating on %d files in %d invalidated targets in %d ' \
'target partitions' % \
(num_invalid_sources, num_invalid_targets, num_invalid_partitions))
# Yield the result, and then mark the targets as up to date.
yield invalidation_check
if not self.dry_run:
for vt in invalidation_check.invalid_vts:
vt.update() # In case the caller doesn't update.