| Commit message (Collapse) | Author | Age |
|
|
|
|
|
|
| |
The option filters proto dependency can be removed from the OptionsParser. This is in response to option parser users that want to avoid the bazel-internal proto file in their dependencies.
RELNOTES: None.
PiperOrigin-RevId: 162249778
|
|
|
|
|
|
|
|
|
|
| |
Whenever we report an action in the build event protocol that has
a configuration associated with it, report the configuration as
well in the protocol (and not only the checksum, if the configuration
is not that of one of the top-level configured targets).
Change-Id: I9b085d5381b3c3509b4f3b99c8a77bc8fba6abfe
PiperOrigin-RevId: 161789745
|
|
|
|
|
|
|
|
|
|
|
| |
From a BEP point of view, the only interface of a configuration we care
about is its BuildEvent structure. Represent it as such, so that we
can move this class to the rest of the buildeventstream module. This
is a prerequisite for ActionOwners refering to configurations in the
BEP.
Change-Id: I6d1c1bf2951aac91607e83cad664553cd6620df8
PiperOrigin-RevId: 161510049
|
|
|
|
|
|
|
|
|
|
|
| |
Set an additional flag in the last BEP message to indicate the end
of the proto sequence. While the last event is implicit from the
child-relation (the sequence of protos is finished, if at least one
event is seen and all announced events have occurred in the stream),
an explicit marking of the last event simplifies processing.
Change-Id: I6554476f975dc9e52fdb27fb3221bd27f6097ed9
PiperOrigin-RevId: 161377092
|
|
|
|
|
|
|
|
|
|
|
|
| |
This is a legacy dependency on the configuration transition table, which is
only needed for static configurations. Dynamic configurations didn't actually
use anything in that table: this was just a convenience interface that could
have equally been defined somewhere else. So this cl defines it somewhere else.
There's still one last dependency: Transitions.configurationHook. We'll tackle
that in a followup cl.
PiperOrigin-RevId: 161141650
|
|
|
|
|
|
|
|
| |
testing.
Unlike in the production flags, these flags are only used for internal testing. Tagged them as NO_OP instead of the default UNKNOWN to make it clear that we do not need these to become properly tagged.
PiperOrigin-RevId: 160526472
|
|
|
|
|
|
|
| |
This lets us change what it expands based on the argument passed to the flag.
RELNOTES: Allows flags that expand to take values.
PiperOrigin-RevId: 160298412
|
|
|
|
|
|
| |
Leave the category for now as the generated docs still do not use the new categorization.
PiperOrigin-RevId: 160290297
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Make the BuildEventStreamer capable of handling the flush() method, even
before the initial build event is seen. This happens, if a lot (more than
10240 bytes) of output is produced before the BuildStartingEvent is generated-
This is not unlikely, as the whole option processing happens before the build
starts.
As the promise of flush() is that the OutErrProvider is cleared, and that all
output between two flush() get into separate events, numbered in order, we have
to temporarily store a list of stdout/stderr String pairs.
RELNOTES: None.
PiperOrigin-RevId: 160137478
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The code previously threw StringIndexOutOfBoundsException if the client
env contained just a variable name with no '=' or value.
Fixed #3196.
Change-Id: I5afcaa398ab2e8bacc709445f50ba363659cadbb
Closes #3197.
Change-Id: I5afcaa398ab2e8bacc709445f50ba363659cadbb
PiperOrigin-RevId: 159222809
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
After having open sourced the Build Event Service code, there is no more need
to have two separate bazel modules that both create a BuildEventStreamer.
This change merges the BuildEventStreamerModule logic into the
BuildEventServiceModule.
DELTA=677 (330 added, 316 deleted, 31 changed)
DELTA_BY_EXTENSION=java=293,oss=32
RELNOTES: None.
PiperOrigin-RevId: 158960687
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
It was originally included in runtime due to external dependencies, and
a desire to keep the options parser a general options library. These
dependencies have been or will be removed, and there are plenty of other
general flag libraries.
InvocationPolicy is fundamentally acting on the properties of this
specific OptionsParser and needs proper access to it for the proper
solution to a number of existing bugs, which means having access to
things that should be package private.
PiperOrigin-RevId: 158523111
|
|
|
|
|
| |
RELNOTES: None
PiperOrigin-RevId: 158279811
|
|
|
|
|
|
|
|
|
|
|
| |
BlazeCommand.editOptions is currently called fairly late in the startup
process, so it must be restrictive in what it does, as any change to the
options can potentially introduce inconsistencies between different parts of
Bazel. Removing the CommandEnvironment reduces the amount of damage it can do,
and may allow us to move the call earlier in the startup process (maybe even
to a point where the CommandEnvironment does not exist yet).
PiperOrigin-RevId: 158133862
|
|
|
|
|
|
| |
RELNOTES: None.
PiperOrigin-RevId: 157446717
|
|
|
|
|
|
|
|
| |
We initially didn't consider that multiple BuildEventStreamer instances might
exist during a build.
RELNOTES: None.
PiperOrigin-RevId: 157385069
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
*** Reason for rollback ***
Roll forward of directory name change
*** Original change description ***
Automated g4 rollback of commit 1d9e1ac90197b1d3d7b137ba3c1ada67bb9ba31b.
*** Reason for rollback ***
Breaks //src/test/shell/integration:force_delete_output_test
*** Original change description ***
Symlink output directories to the correct directory name
If the workspace directory is /path/to/my/proj and the name in the WORKSPACE
file is "floop", this will symlink the output directories to
output_base/execroot/floop instead of output_base/execroot/proj.
More prep for #1262, fixes #1681.
PiperOrigin-RevId: 156892980
|
|
|
|
|
|
|
|
|
|
| |
In preparation to support multi-configuration builds, provide infrastructure
allowing build events to reference BuildConfigurations. The streaming mechanism
will ensure that build configurations are introduced in the stream before being
referenced for the first time.
Change-Id: I6b96fbebc76a05eff4f75a07e8a9cfbcd57f9c22
PiperOrigin-RevId: 155368666
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
*** Reason for rollback ***
Filter out conflicting flag policies for applicable commands. Only enforce the filtered, applicable policy.
*** Original change description ***
Automated g4 rollback of commit aa7f9307636d38cbb93a03acac8f4c59adfa0ee8.
*** Reason for rollback ***
Broke --experimental_inmemory_dotd_files
*** Original change description ***
Filter out conflicting flag policies before invocation policy enforcement.
This is to minimize the likelihood of obscure policy conflict. Now, the last policy on a flag (after policy expansion) will be the only one in the "canonical" invocation policy. There should be no reason for explicitly setting multiple policies on a single flag, but if an expansion flag is policy'd and one of its children has a more specific policy on it, make sure that the policy on the child flag is after the policy on the expansion flag.
Note that this restriction (only the last policy gets applied) also applies for repeatable flags. Make sure all values being set to a repeatable flag are set in a single SetValue operation, with multiple flagValues set.
PiperOrigin-RevId: 154969189
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Extend the build-event streamer to also report about stdout/stderr,
if provided. This information is reported in the progress events. At
the moment, we only report stdout/stderr in progress events we send
anyway, but the interface is generic enough that we could add time-based
reporting later, if needed. Also note, that at the end of the build, we
report the final progress event, so that all stdout/stderr generated before
the build-complete event get also reported in the build-event protocol.
Change-Id: If5dbd59c151edbce02d0a9b2e5938b63c0a5dc58
PiperOrigin-RevId: 154811110
|
|
|
|
|
|
|
|
|
| |
*** Reason for rollback ***
Broke --experimental_inmemory_dotd_files
RELNOTES: None
PiperOrigin-RevId: 154477949
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
*** Reason for rollback ***
Rollforward with fix for test flakiness. BEP transport closed
events are delivered via their own threadpool and thus might
not have been sent immediately. BuildEventStreamerTest#testSimpleStream
now waits for a bit until the event has been delivered. I ran the test
with --runs_per_test=1000 several times and had no further failures.
*** Original change description ***
Automated g4 rollback of commit 3d596d63f883fff56001ed7b2e5cf51dba45f082.
*** Reason for rollback ***
Made BuildEventStreamerTest#testSimpleStream 3% flaky based on --runs_per_test=1000.
RELNOTES: None
PiperOrigin-RevId: 154170833
|
|
|
|
|
|
|
|
| |
*** Reason for rollback ***
Made BuildEventStreamerTest#testSimpleStream 3% flaky based on --runs_per_test=1000.
RELNOTES: None
PiperOrigin-RevId: 154090559
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The BEP (build event protocol) upload may at times take longer than the build
itself. This is especially true for cached builds, where there is little build
work, but the protocol still needs to be uploaded. In this case the bazel UI
will now display that it's waiting for BEP upload, both in the current and the
new experimental UI (--experimental_ui).
When executing a run command, blaze waits for the BEP upload to finish before
it runs the target.
Major Modifications:
- The BuildEventTransport interface now also has a name() method that returns
a string. When waiting for a transport to finish the BEP upload, this string
is displayed in the UI.
- The BuildEventStreamer now emits two new events
AnnounceBuildEventTransportsEvent and BuildEventTransportClosed on the event
bus. This is how the experimental UI is informed about BEP upload progress.
RELNOTES: None
PiperOrigin-RevId: 154052401
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This is to minimize the likelihood of obscure policy conflict. Now, the last policy on
a flag (after policy expansion) will be the only one in the "canonical" invocation
policy. There should be no reason for explicitly setting multiple policies on a single
flag, but if an expansion flag is policy'd and one of its children has a more specific
policy on it, make sure that the policy on the child flag is after the policy on the
expansion flag.
Note that this restriction (only the last policy gets applied) also applies for repeatable flags. Make sure all values being set to a repeatable flag are set in a single SetValue operation, with multiple flagValues set.
PiperOrigin-RevId: 153584034
|
|
|
|
|
|
|
|
| |
There's no reason an expansion flag should not expand to multiple values for a repeatable flag (a flag with allowMultiple set to true.) If this expansion flag is set in a SetValue policy, group its repeatable subflags into a single SetValue per subflag.
For an overridable SetValue policy on an expansion, any repeatable flag that it expands to should append its value, and not override the user's original values.
PiperOrigin-RevId: 153233784
|
|
|
|
|
|
| |
By the time the invocation policy gets enforced, the expansion flag should have been
expanded, and policy should act as on any other expansion flag.
PiperOrigin-RevId: 152692831
|
|
|
|
|
|
|
|
|
|
| |
Extend the functionality of the BuildEventStreamer to report those parts
of NestedSets of Artifacts not reported earlier. In this way, duplicate
reporting can be avoided, without the events themselves having to know
which artifacts are known already.
Change-Id: Ia959c28c440301860eac57ea5d9a712c0d49ebdf
PiperOrigin-RevId: 152497672
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Better specify the BuildEventTransport interface. Besides clarifying threading
and blocking issues, this change also clarifies error handling/reporting.
After several discussions we concluded that the BuildEventTransport interface
should not provide error reporting / handling facilities, as there is not much
bazel could do with this information. Instead, a transport may decide for
itself if an error is fatal and abort the build or if an error should be logged
to the user's terminal or if it should be ignored.
Furthermore, changing the close() method lays the groundwork for an upcoming
change that will report the transport shutdown status to the user command
line.
PiperOrigin-RevId: 152408938
|
|
|
|
|
|
|
|
|
|
|
|
| |
'create' method.
This paves the way for changing PathFragment to e.g. an abstract class with multiple subclasses. This way we can split out the windows-specific stuff into one of these concrete classes, making the code more readable and also saving memory (since the shallow heap size of the NonWindowsPathFragment subclass will hopefully be smaller than that of the current PathFragment).
This also lets us pursue gc churn optimizations. We can now do interning in PathFragment#create and can also get rid of unnecessary intermediate PathFragment allocations.
RELNOTES: None
PiperOrigin-RevId: 152145768
|
|
|
|
|
|
|
|
|
|
|
| |
Change the BuildEvent interface to accept a generic class of converters.
In this way, we won't have to change it again in the future, once more
converters are needed. In fact, a new converter is needed right now (will
be added in a follow-up patch) to allow build events to know the name of
named artifact groups already reported in the stream.
Change-Id: Ibb32ea5fff361e21bcf2d34818d8351a1da7a2e3
PiperOrigin-RevId: 152131870
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
*** Reason for rollback ***
Breaks //src/test/shell/integration:force_delete_output_test
*** Original change description ***
Symlink output directories to the correct directory name
If the workspace directory is /path/to/my/proj and the name in the WORKSPACE
file is "floop", this will symlink the output directories to
output_base/execroot/floop instead of output_base/execroot/proj.
More prep for #1262, fixes #1681.
PiperOrigin-RevId: 152126545
|
|
|
|
|
|
|
|
|
|
|
|
| |
--incompatible_* flags
Note that if a developer adds a poorly-formatted incompatible change @Option, constructing an OptionsParser will now fail with an unchecked exception. This can cause some unit tests to fail in unexpected ways, but the developer should see an appropriate error message when the server starts up.
To be added: A separate integration test that ensures the expansions of --all_incompatible_changes don't clobber each other.
RELNOTES: None
PiperOrigin-RevId: 151858287
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
--noexperimental_enable_critical_path_profiling flags are all specified, then Bazel will delete Actions from ActionLookupValues as they are executed in order to save memory.
Because an already-run action may output an artifact that is only requested later in the build, we need to maintain a way for the artifact to look up the action. But in most cases we don't need to keep the action itself, just its output metadata.
Some actions unfortunately are needed post-execution, and so we special-case them.
Also includes dependency change with description:
Move action out of key. This keeps action references from polluting the graph -- actions are just stored in one SkyValue, instead of being present in SkyKeys.
This does mean additional memory used: we have a separate ActionLookupData object per Action executed. That may reach ~24M for million-action builds.
PiperOrigin-RevId: 151756383
|
|
|
|
|
|
|
|
| |
For SetValue and UseDefault policies on expansion flags or flags with implicitRequirements, expand the policy into policy for each of its sub-flags. For SetValue, this addresses the issue with policies on expansion flags with overridable=true not actually letting user flags overrride the expansion. For UseDefault, this formalizes the behavior where UseDefault will wipe all user-provided flags that expand from a banned expansion flag, and will allow later work to guarantee that a later policy can override the expansion policy's subflags.
Since expansion flags do not have value, break if the invocation policy uses AllowValue or DisallowValue on an expansion flag.
PiperOrigin-RevId: 151718539
|
|
|
|
|
|
|
|
|
|
| |
If the workspace directory is /path/to/my/proj and the name in the WORKSPACE
file is "floop", this will symlink the output directories to
output_base/execroot/floop instead of output_base/execroot/proj.
More prep for #1262, fixes #1681.
PiperOrigin-RevId: 151712384
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
It's possible for cases to happen where the BlazeCommandDispatcher
fails before starting the WaitForCompletionCommand. In these cases,
the test's command state handoff will never take place because no
command state will be created to be handed off (obviously).
This means that runIn will wait forever.
Instead, if the command completes and the command state has still
not been handed off, set an exception into the command state so
that the test will fail without further troubles.
RELNOTES: None.
PiperOrigin-RevId: 151495494
|
|
|
|
|
|
| |
The encapsulation was off, since the format in which we accept policy has nothing to do with how we then enforce the policy.
PiperOrigin-RevId: 151471747
|
|
|
|
|
|
|
|
| |
It was large and unwieldy - keep common logic in the original file but split out the tests into groups by invocation policy operation type.
--
PiperOrigin-RevId: 150759733
MOS_MIGRATED_REVID=150759733
|
|
|
|
|
|
|
|
| |
Remove magic constants and convert to Truth.
--
PiperOrigin-RevId: 150692720
MOS_MIGRATED_REVID=150692720
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Always initialize the set of postedEvents. Also, record the initial event
as posted. In this way, the build event streamer can also handle events that
should be posted after the BuildStartedEvent but are told the streamer about
earlier.
--
Change-Id: Ic796f2434434819e6c62633755411527a9dcb7e0
Reviewed-on: https://cr.bazel.build/9451
PiperOrigin-RevId: 150620387
MOS_MIGRATED_REVID=150620387
|
|
|
|
|
|
| |
--
PiperOrigin-RevId: 150051360
MOS_MIGRATED_REVID=150051360
|
|
|
|
|
|
| |
--
PiperOrigin-RevId: 149418372
MOS_MIGRATED_REVID=149418372
|
|
|
|
|
|
|
|
|
|
|
|
| |
In the experimental UI, to be able to report appropriately about ongoing
downloads, we need to track their state, as updated by the DownloadProgressEvents.
Do so. Also report on ongoing downloads in the progress bar.
--
Change-Id: I668e963cd6da85ec598b23724066d366d465271f
Reviewed-on: https://cr.bazel.build/9114
PiperOrigin-RevId: 148899297
MOS_MIGRATED_REVID=148899297
|
|
|
|
|
|
|
|
|
|
|
|
| |
The build event protocol now emits a BuildFinished event at the end of a build. The event is a child of the BuildStarted event.
The code changes were larger than expected, due to some refactoring in BuildEventStreamer. This was necessary as the BuildCompleteEvent now implements the BuildEvent interface and Guava's EventBus always invokes the most specialized subscriber method. Thus, the buildEvent(BuildEvent) and buildCompleted(BuildCompletedEvent) methods had to be merged.
--
Change-Id: Id0c2bd7220dc8ce03128b7126587e212ee8ce836
Reviewed-on: https://cr.bazel.build/9053
PiperOrigin-RevId: 148788582
MOS_MIGRATED_REVID=148788582
|
|
|
|
|
|
|
|
|
|
|
| |
Bazel-created files (like log files of test runs) are internally reported
as Paths. However, this is not always the most useful representation of the
location of that artifact for a consumer of build events. Therefore, support
a mapping of paths to more useful URIs.
--
PiperOrigin-RevId: 144843525
MOS_MIGRATED_REVID=144843525
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
which have converters that return lists. The problem was that the policy might,
say, disallow values ["a", "b"], and a flag --foo might have a converter which
takes a string and splits it by commas to produce a list. The options parser
would apply the converter to --foo=a,b to produce ["a", "b"], but invocation
policy would compare each element of the policy to the list itself, which will
never work.
--
PiperOrigin-RevId: 142297177
MOS_MIGRATED_REVID=142297177
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
If an Aspect registered an action that an extra-action is
shadowing, its name is used when creating the extra-action's ID and
name.
Since recently, an aspect can see other aspects applied to the same
target. This CL record the names of other aspects applied to the target
as well, disambiguating the action owners.
--
PiperOrigin-RevId: 142264153
MOS_MIGRATED_REVID=142264153
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
In the build event protocol, we promise to respect certain ordering
constraints that come naturally by causality anyway (e.g., the root
cause of a failure has to come before the failure). However, in the
way the call structure is organized, events might occur on the event
bus in wrong order. So allow events to specify that order and make the
streamer honor it.
--
Change-Id: I1fbe9b93681f0ddfa35d9d00d5d1b02103da38a5
Reviewed-on: https://bazel-review.googlesource.com/#/c/7370
MOS_MIGRATED_REVID=139774946
|
|
|
|
|
|
|
| |
its buffer.
--
MOS_MIGRATED_REVID=139771073
|