| Age | Commit message (Collapse) | Author |
|---|
|
|
|
(cherry-picked from 1e5cc37df36204474cedef68486ba8b29492d259)
|
|
... and also perform conflict check on the normalized paths. Still,
the the output of the "ACTION" funtion be keyed by the representation
of the paths as originally described, to allow the author of a rule
to use non-normalized paths as well.
(cherry-picked from 74a5fbc9644eda7f105d2d2c672a41fa5c34c6a8)
|
|
Due to the specifics of how CLI11 library handles the --help
argument and the fact that just-mr uses its own set of exit codes,
we were correctly printing the help text, but falsely returning a
non-zero exit code as result.
This is now fixed.
(cherry-picked form faa8fc156ed8314db23a2d1612975ee745a50ade)
|
|
During multithreaded splicing, the main process can be forked (inheriting open file descriptors). In this case, the executable file saved using hardlinking becomes inaccessible. To prevent this, executables must be stored as copies made in a child process.
(cherry-picked from a336b835ee0831ba5fe2f4e8f9b1773eff68416b)
|
|
After successful build of an export target target a cache entry is
written. As those export targets have a dependency structure, care
is taken to write them in correct order.
Writing a cache entry for an export target requires knowledge of
the analysis result of that target. Originally, all targets used
to be analysed locally, so that information was available for
every target that was not read from the local target cache already.
However, with the introduction of serve endpoints, it can happen
that a target was analyzed locally, but it depends on targets
obtained from serve. In this case, we have to refrain from writing
a target-level cache entry to keep the the consistency invariants
of the target-level cache.
(cherry-picked from f63d3df866a2d75b694c8c484b21fbcdccd58acd)
|
|
...fixing potentially dangerous code (evaluation order is unspecified).
(cherry-picked from b91f39d2429704859e799bff2554fc40636438f9)
|
|
|
|
|
|
When populating the GitTree instance stored in a Git tree-type
FileRoot with the ignore-special flag set, the GitTree instance
would be created with an empty raw_id_ field, signaling that some
of the entries might have been skipped and thus the root tree id
is not anymore in a one-to-one correspondence with the stored list
of entries.
This however caused FileRoot instances with missing tree id
information. This commit fixes the issue by always storing the
raw_id_ field as the root id of the Git tree, as well as clarifying
the relationship between this field and the ignore_special_ flag,
including refactoring the tree id getters.
(cherry picked from 06336198a38ffaec6c35df34580391579a7a7b22)
|
|
If the serve endpoint reports an internal error, local builds
should not continue and the error message should be provided.
Similarly, if the serve endpoint promises the target cache value to
be in remote CAS, but the client cannot find or process it as
needed, then a local build again should not continue and the reason
be provided as an error message.
(cherry picked from ab0a06871391a84c9b904bc9f2cd74ee1e8903ea)
|
|
|
|
|
|
Compared to the command line, the environment usually is quite
short. So including it in messages reporting about commands does
not introduce a lot of additional noise. However, knowing the
environment can help understanding an error message. Therefore, it
seems a good trade off to include it. Do so.
|
|
|
|
|
|
The gsl-lite implementation is slightly more picky in terms of
type conversions and constness resolution in initializers,
therefore small changes were needed.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
To avoid lock contention, our task system queues tasks in several
queues. In order to know when there is no more work to be done, a
global counter keeps track of the total work. Here, it is important
that this counter be incremented before a value is added to the
queue, for otherwise some other thread could pick it up and decrement
the total work load, hence getting us in a state where the counter
underestimates the total amount of work to be done. Fix the order
of those operations.
|
|
... by taking the correct map to report which targets failed
to be analysed and also report about rules that failed to be
analysed.
|
|
Also aligning comments in server-side headers with the proto file
documentation.
|
|
Also updates the tests and all relevant documentation accordingly.
|
|
This will make it available also to just.
|
|
... of an absent target. Here, the request is given by repository
root and filename; so the filename is to be taken relative to the
root, i.e., we have to prefix the targets-file name with the module.
|
|
The result of the analysis is a JSON object containing the keys
`"artifacts"`, `"runfiles"`, and `"provides"`. This JSON object, by
default, is logged. However, it might be useful to process the data
contained in it, while, for example, developing new rules.
This patch adds a new command line option (`--dump-result`), reserved
to the subcommand `analyse`, to dump the analysis result to the given
file or stdout (if `-` is given).
|
|
While this can already be expressed by an "if" statement, having
a dedicated function for logical negation makes some expressions
more readable.
|
|
... using, also for the "then" branch, the empty list as default.
In this way, this statement not only more symmetric, but also
allows shorter representations of some typical expressions.
|
|
Lists are somtimes used in configurations as replacement for tuples.
Providing length gives an easy way to detect usage errors.
|
|
|
|
For setup and update subcommands, the consumer lambdas for their
respective async maps are non-trivial (i.e., do more than keeping
the map values) and operate on ungaurded JSON objects, thus it is
better to guard them against any uncaught exceptions.
|
|
When taking over roots from the input repository description, the
JSON value was not properly checked for validity (i.e., if it is of
type string), resulting in an unhandled exception being thrown. The
issue is fixed with a proper type check.
|
|
... as also just describe can communicate to grpc endpoints.
|
|
During compactification, invalid entries must be deleted.
|
|
|
|
Commit f5f9be5bc07b16807aceac86fba9212e3889762a changed from
tracking progress by cache key to tracking by configured target;
however, the absent-target map was forgotten. Fix this while also
switching to the shortend export-target representation introduced
in the previous commit.
|
|
As we always analyse export targets in their canonical configuration (and
do count in the progress the mapping from obtained configuraiton
to canonical one), the shortend name (i.e., the representation
with null values dropped form the configuration) is still a unique
representation of the target. Use this in progress reporting to
simplify reading the progress sample.
|
|
|
|
... using the same mechanism that just uses. As just-mr now also
does quite some grpc calls, e.g., fetching distribution files from
the remote-execution endpoint and negociating roots with the serve
endpoint, it may now also be affected by a bad network or a service
temporarily be unavailable.
|
|
... to simplify reuse.
|
|
During garbage collection split and remove from the storage every entry that is larger than a threshold.
|
|
During garbage collection remove from the storage every entry that has the large entry.
|
|
|
|
|
|
|
|
|
