| Age | Commit message (Collapse) | Author |
|---|
|
Currently, the implementations of the split and splice operation are both
hidden behind the Bazel API implementation. This was sufficient to implement
splitting at the server and splicing at the client. In order to support the
other direction of splitting at the client and splicing at the server while
reusing their implementations, the code needs to be refactored. First, the
functionality of split and splice are explicitly exposed at the general
execution API interface and implemented in the sub APIs. Second, the
implementations of split and splice are factored into a separate utils class.
|
|
|
|
|
|
|
|
|
|
Add an end-to-end test verifying that we report enough useful
information in case of an action failing.
|
|
... if provided. This might help users to find the correct place
in their code base causing the action to fail.
|
|
... i.e., the values for "mirror" and "inherit env"
|
|
When importing a repository via just-import-git, allow to
specify the value for the "inherit env" property for the
repository just being imported.
|
|
|
|
When fetching git repositories, just-mr routinely shells out to
git. In this case, allow the user to specify via "inherit env",
which environment variables from the host environment should be
made available in this action. Typical variables to inherit are
ones providing credentials, like SSH_AUTH_SOCK. As the repository
description specifies the commit that will be taken, and hence the
resulting tree, correctness is not affected by the environement
leaking in here.
|
|
At some point we have to decide if a given git repository URL is
a path. So far, we assumed that anything not starting with ssh://,
http://, or https:// is a path. This ignores the facts that
- the file:// scheme, while referring to a file, does not denote
a relative path starting file://,
- the [user@]host:path scheme is not a path on the local machine,
- there exist the URL schemes git://, ftp://, and ftps://, and
- future extension might add additional schemes.
To also correctly handle new schemes that git might add (which we
indeed can handle, as we simply shell out to the git binary), we
reverse the approach: we give the user the means to unambigiously
specify that they refer to a path on the local machine, by either
- using the file:// scheme,
- providing an absolute path starting with /, or
- providing a relative path starting with ./
All other schemes will not be modified. The file scheme, as well
as the git://, http://, and https:// scheme, are handled interally
using libgit2; all others are passed on to git in an unmodified form.
|
|
We ensure that for each export target to be written to the target
cache all its implied export targets are written to the target
cache first. This ensures that the target cache maintains its
consistency at all times with respect to export target
dependencies.
|
|
|
|
As it is common to use ArtifactDigest objects with a default value
of 0, the std::hash implementation should not take the size into
account. The is_tree member however should be kept in as it always
needs to have a value and has no sensible default.
|
|
|
|
|
|
|
|
... of all produced binaries, including the intermediate
ones: protoc and grpc_cpp_plugin.
|
|
... glibc provides synchronization stubs for single-threaded
environments as weak symobls. When linking pthreads, these
weak symbols must be replaced by the strong symbols provided
by the pthread library. For dynamically linking pthreads,
this is done automatically. However, to support this for
static linking, we must ensure to link the whole archive.
|
|
... as any unguarded access to non-const members of the same
shared_ptr instance require the use of `atomic_load` and
`atomic_store`.
|
|
... not that everyone has updated to 1.2 or later we can use
the built-in expressions "reverse" and "set".
|
|
... to provide an informative error message on how a rule is related
to a particular import and, in particularly, at which expression
a problem with the import occurred.
While there, also improve the message in the other error case to
follow our standard line-breaking scheme.
|
|
|
|
... in order to not rely on the default launcher to pull it in.
|
|
... to not rely on env implicitly pulling in echo to PATH.
|
|
... in order to not assume echo to be on the standard search
path pulled in by env.
|
|
For a user `just install-cas` will show an entry without revealing
where it found it---as it is content-addressable, it does not matter.
Therefore, verify that accessing paths of a tree object also works
regardless of where the tree is stored.
|
|
|
|
By showing the full entity name and also adding the usual
newline character after every "While ..." clause.
|
|
|
|
... instead of erroring on missing file. In this way, whenever a
rule or expression from an absent root would have to be read, we
get a meaningful error message and not a complaint about a file
not being there.
|
|
|
|
... instead of hard-coding it in the example graphs. In this way,
we can honor a provided PATH.
|
|
|
|
... to make tests pass also on systems where /bin/sh does not,
by default, pull in a path to "the standard tools".
|
|
... and only let test do the check on the final resulting boolean,
where the string representation is canonical. In this way, we avoid
having to rely on the string representation of numbers, where, e.g.,
1 and 1.0 are equally valid representations of the same number.
|
|
... in the local launcher for the execution end point. In this way,
the tests also work on systems where sh does not pull in enough
paths to have all the "usual" tools available.
To also get a correct path on systems relying on sh pulling in
enough paths, take what sh, launched with the default launcher,
would believe the path is.
|
|
... for test actions, by setting an appropriate local launcher. In
this way, the tests can also be run on systems where sh does not
pull in enough paths to have all the "usual" tools available.
|
|
As we patch that API, it also can implicitly depend on the
toolchain (and hence its config), if we use a compiled implementation
of patch. Making the TOOLCHAIN_CONFIG a flexible part of the config
allow to, again, build static binaries as usual.
|
|
|
|
... as, for absent repositories, we need to get the description from
the serve end point. As a consequence, also support -r and --compatible,
as the remote-execution endpoint needs to fit with the one for the
serve endpoint.
|
|
... as even this command now adds entries to CAS, e.g., the shard
for the target-level cache. Also, this command block gc by keeping
a lock (also in the local build root).
|
|
Also cleans up the logging when parsing the serve service
configuration file.
|
|
As the serve service always has an associated remote-execution
endpoint, when just-mr gets passed only a serve endpoint address it
will now default to implying a remote-execution endpoint exists too
at that address. Additionally, now the implementation of the
--backup-to-remote option more clearly shows that it can only be
done in native mode.
|
|
The requests to retrieve the tree of a commit, archive, or distdir
also set up those trees in a way that the serve endpoint can later
build against them, besides allowing just-mr to set up roots
locally. Therefore, if the witnessing entity (Git commit, content
blob, or distdir, respectively) is known to the serve endpoint,
then failing to set up the root tree there should result in a
failure also of the just-mr setup on the client side.
|
|
For archives and Git repositories we should ensure that not finding
the witnessing entity (archive content blob or Git commit,
respectively) results in a distinct status in the response to a
request that sets up roots on the serve endpoint. This will allow
just-mr to better handle its interaction with the serve endpoint.
|
|
A change made to this test was wrongly setting the same path both
as build root for the serve instance and for the just-mr instances
being tested, resulting in the build root of the serve instance
being inadvertently removed during the test. This commit fixes the
issue by giving the serve endpoint its own build root path.
|
|
|
|
While just-mr does not use remote-execution properties, it is
still useful to have those as a separate entry in the rc file. With
rc-file delegation, this gives committed rc files an easy way to
specify the image to be used without having to set all the remaining
arguments for the various just subcommands in "just args".
|
