| Age | Commit message (Collapse) | Author |
|---|
|
...in creating Git tree from filesystem directory.
|
|
|
|
|
|
...in methods that should not report at error level themselves,
but let this be handled by its callers.
While there, remove an unneeded path manipulation in a defferent
set of log messages.
|
|
...while also removing some unneeded one.
Do not implicitly trust that the third-party code called in these
methods is non-throwing and instead properly handle any exception
that might arise. Also remove the specifiers from some anonymous
namespace methods where a try-catch would be overkill and let
their callers handle any exceptions instead.
|
|
...and fix inconsistent capitalisation.
|
|
Do not implicitly trust that the third-party code called in these
methods is non-throwing and instead properly handle any exception
that might arise.
|
|
|
|
... to include at least all pages referenced throughout the man page.
|
|
|
|
|
|
... which should not do any symlink checks in compatible
mode.
|
|
|
|
This ensures that any entries that the standard remote execution
protocol accepts but are invalid in justbuild, i.e., upwards
symlinks, are rejected.
For this purpose, do not fail in the action response instances,
just perform the check there, as all required information is
available, and set a flag that the executor can check as needed.
|
|
While in practice a failure to populate the fields of a response
happens once per invocation, as it will trigger a failure of the
execution, from an algorithmic standpoint the flag to mark a
successful population of the response fields should only be set on
actual success.
Fix this.
|
|
...and remove specifiers from methods that might throw in
unexpected ways. By doing this, balance the need to avoid wrongly
silencing exception sources during execution with reducing the
amount of try-catch blocks.
|
|
This will check if directories contain upwards symlinks.
|
|
The execution server itself should not consider anything special in
setting the response message to the client, instead let the
underlying API fail or not during collection.
|
|
|
|
In particular, ensure that Git roots check for, e.g., upwards
symlinks, before returning blobs and trees. To ensure that only the
bear minimum extra work is performed for this purpose, Git roots
now keep also the root's GitTreeEntry as a field, allowing the
validity check of root source trees to take place only once and
only if required.
|
|
|
|
|
|
...with respect to rejecting invalid entries such as upwards
symlinks. Also ensure that valid trees are only checked once by
remebering known valid tress though marker files in local storage.
|
|
...whenever it is given access to a Git repository.
The referenced storage config needs to outlive the repository
config instance.
|
|
...through marker files kept in storage under generation regime.
These can be used to allow valid source trees, i.e., those free of
upwards symlinks, to be cached in a persistent manner over multiple
builds.
|
|
This is useful when the caller already knows that the tree to look
up is valid, and thus the extra check step can be safely skipped.
|
|
This removes a scenario where otherwise successful (exit code 0)
calls to just and just-mr would result in an error-level log
message.
|
|
|
|
Match behaviour of reading trees, which always checks for invalid
entries, also for reading blobs.
|
|
This allows individual blobs read to be checked, e.g., for upwards
symlinks, also when not part of a tree, which performs such a
validation for its entries during its parsing into a GitTree.
|
|
|
|
|
|
|
|
Often, the simple number, e.g., for failed actions can already give
valuable information. For example, when investigating flakiness,
the count of failed actions already gives a hint whether a change
increased or decreased flakiness which can be valuable before even
investigating if the nature of the failure has changed. As we have
that information available and an additional number in a heading
does not clutter the page too much, let's just show it.
|
|
... to improve readability of target files.
|
|
... and do mark artifacts internally as synchronized. First all all,
we will abort anyway, to the entry won't even be read and, secondly
it is not necessarily true that the artifact is synchronized.
|
|
|
|
|
|
If the main repository is marked absent, warn if during the
dependency closure computation any non-content-fixed repositories
are reached, i.e., any "file"-type repositories that are neither
implicitly nor explicitly marked "to_git". Also warn if the main
repository itself is marked absent but is not content fixed.
Add small test checking that the new warning is produced.
|
|
When backing up target-cache entries we use parallelism at two
dimensions, the independent cache entries and for each entry we
retrieve the artifacts in parallel. If for each dimension we use
the full amount of parallelism allowed, that gives a number of
threads up to the square of the amount of parallelism specified by
the user. Therefore, use in each dimension only the square root
of the allowed parallelism keeping the total parallelism (up to
rounding) within the specified range.
|
|
Add a test testing the serve instance implicit to the "with serve"
rule by asking it to build a target that creates an export target
where the output is a deep tree.
|
|
Add a test asking serve to provide a target where both, the dependency
as well as the action graph has both a wide node (a target with a
large number of direct dependencies) as well as a deep node (a node
where the chain of dependencies is long). That target is requested
serveral times in parallel to verify that such targets can also be
served under load.
The process actually tested is the `just serve` process implicit
to the "with serve" rule.
|
|
Dropping unnecessary "arguments_config" as well as evaluating
trivial computations.
|
|
|
|
As the end of the build can be significantly ahead of the end of
the incvocation, a new time stamp was added to the profile. Use
this new time stamp in the invocation server when determining the
wall-clock time of the build.
|
|
... which might be quite ahead of the end time of the invocation if
writing out of the action graph delays the end of the invocation.
|
|
Also show only the build time in the invocations overview, as only
for the build phase detailled timing information will be available
in the log of a particular information.
|
|
...in all return paths, including in reporting caught exceptions.
In this way give the opportunity for any calling AsyncMap to
receive an expected fatal logger call on failure and thus be able
to shut down gracefully. This is in line with the AsyncMap design,
where the loggers are assumed to be safe to call by a consumer.
|
|
|
|
|
