- Jan 09, 2024
-
-
Georges Racinet authored
-
- Jan 07, 2024
-
-
Georges Racinet authored
- using the node id rather than the tag in `mercurial.rev`, to dodge the usual problems with tags - had to pin down the new `toml` dependency and the `serde_spanned` it brings because MSRV of serde_spanned 0.6.5 is 1.67.0 - a few into() were needed due to the introduction of `UncheckedRevision`. Mostly because we get `Revision` from nodemap calls. Perhaps this can be simplified later on. This is expected to be 0-cost anyway.
-
- Jan 03, 2024
-
-
Georges Racinet authored
Structured errors in `GetTreeEntries` will be a larger effort, should be done this week, though.
-
- Nov 10, 2023
-
-
Georges Racinet authored
In the case of unknown storage, we find it useful to repeat the requested name, hence we simply stop comparing exactly the details. In the case of the missing repository field, the new upstream message is clearly better, so we adopt it.
-
- Sep 24, 2023
-
-
Georges Racinet authored
It turns out that Gitaly returns an empty response on an empty repository, whatever the value stored as default branch might be. In case the default branch points to a branch that does not exist, it returns the first branch it can find. This latter case cannot happen with the state maintainer provided by py-heptapod. This can be important for various mechanisms to actually set the default branch. We're simply returning an empty response if the default branch is set but its target does not exist. This amounts to the same in all current scenarios. Later on we could return the first GitLab branch if that becomes useful.
-
- Sep 20, 2023
-
-
Georges Racinet authored
Of course this requires an update of the base image, to get a Gitaly past the point when `root_oid` is removed from protocol. We could run locally the Comparison tests with 16.0.0, which is simpler to reference than an exact commit, so we'll try this. The length of the first response in `test_compare_get_tree_entries_pagination` by Gitaly changes, perhaps because each `TreeEntry` is a bit smaller (we've long suspected the streaming logic was based on actual size hints), but we did not check that. Closes #151
-
- Aug 31, 2023
-
-
Georges Racinet authored
Testing the case of `limit=0` and revision unknown uncovered a small difference between HGitaly and Gitaly (RHGitaly was already compliant): HGitaly was returning a response, with an empty list of entries. Decided to change `chunked()` to avoid yielding an empty chunk, but it turned out that `RepositoryService.SearchFilesByName` was actually expected to yield responses with empty lists, hence we hade to make it optional. The check in `CommitService` to avoid empty chunks having become default, we had to remove it to kepp coverage.
-
- Aug 30, 2023
-
-
Georges Racinet authored
This builds on the previously introduced elements: iterators for the three cases, helpers to stream in paginated chunks and activates the Gitaly Comparison tests for this method. The actual call to the iterators look to be identical, but they are not, as monomorphisation will give us three different `stream_get_tree_entries_from_iterator`.
-
Georges Racinet authored
This is the engine behind the non-recursive case of `CommitService.GetTreeEntries`, if `skip_flat_paths` is `false`. Like the Python reference HGitaly implementation, we are interpreting the "flat path" to be the greatest common path of all entries equal or inside the given entry (see `hgitaly.manifest` Python module for details about this)
-
Georges Racinet authored
This is the core engine for one of the three modes of `CommitService.GetTreeEntries`.
-
- Aug 31, 2023
-
-
Georges Racinet authored
This iterator will be the core engine of `CommitService.GetTreeEntries` in the simplest case: non-recursive without flat paths computation. The flat paths computation is not as expensive in the Mercurial case as it is in the Git case, because we are iterating over the entire manifest anyway (a later version might use a binary search to find the starting point of the requested directory, but all files within the directory will have to be scanned), but not doing it enables this simple implementation: namely we can yield top-level directories immediately, and hence to have at most one `TreeEntry` to yield per run of the loop.
-
- Aug 30, 2023
-
-
Georges Racinet authored
This utility method finds the greatest common denominator of two paths, and helps using the remainder in one of them. It takes care of the various edge cases (strict equality, trailing slashes) and will be used several times in the `GetTreeEntries` implementation.
-
Georges Racinet authored
This will be useful to implement the various cases of `GetTreeEntries`. It could also be extended to be used in `ls_path`, but that would require treating the special case where the given path is actually a file. Consider sending this UPSTREAM.
-
Georges Racinet authored
This generic helper in `rhgitaly::streaming` is for the case of streamed responses with `repeated` items, the whole stream being the current page. In all cases known to us, GitLab derives some information from the last item of the last chunk to be the `next_cursor`, hence we introduce a trait for the item to represent that, and use it in the implementation. The first use-case will be the implementation of `CommitService.GetTreeEntries`, but this also makes the likes of `RefService.FindLocalBranches` essentially trivial.
-
Georges Racinet authored
This Rust version of Python `hgitaly.util.chunked` should be useful for many gRPC methods. It provide the means to distinguish the first response, but does not implement the pagination protocol.
-
- Aug 18, 2023
-
-
Georges Racinet authored
Probably more complicated than it should, but it works and can be swapped later to something more straightforward.
-
- Aug 12, 2023
-
-
Georges Racinet authored
Closes #135
-
Georges Racinet authored
This parallels Python's `heptapod.branch.gitlab_branch_ref`.
-
- Aug 07, 2023
-
-
Georges Racinet authored
Adaptation to match changes of Gitaly error details in `CommitService.TreeEntry`
-
- Jul 31, 2023
-
-
Georges Racinet authored
We put the previously introduced building blocks together and validate with the Gitaly Comparison test, with some improvement needed in its expensive fixture left for follow-ups.
-
- Aug 07, 2023
-
-
Georges Racinet authored
There are for now completely identical to methods for blobs, but they could well become different (e.g, different disk caches or whatever).
-
Georges Racinet authored
This will be useful to implement `CommitService.TreeEntry`. There is plenty of room for performance improvement, notably: - the position of the first relevant entry could be found by binary search instead of a linear scan - we could try and clone less But it is possible that this would not be really needed, given the performance and scalability boost given by RHGitaly over HGitaly, before Mercurial manifests start using a more efficient layout, making this code obsolete.
-
Georges Racinet authored
We will soon implement `CommitService.TreeEntry`, which has to behave like `GetBlobs` on one `RevisionPath` in case it turns out the path is a blob.
-
- Jul 31, 2023
-
-
Georges Racinet authored
We split `lookup_blob()` in two functions, so that parts can be reused without reopening changelogs, manifests etc. Notably, the implementation of `TreeEntry` will use `read_blob` in case it is serving a blob and not a tree: in this case the manifest entry has already been found, no need to reopen and rescan anything.
-
- Aug 06, 2023
-
-
Georges Racinet authored
Rust 1.65 is currently the version generally available in GitLab context (upstream usage appeared recently) and it is also able to build tokei 12.1.2, whereas 1.61 cannot. This leads up to fix the following Clippy error (not in 1.61): "you are deriving `PartialEq` and can implement `Eq`".
-
- Jul 31, 2023
-
-
Georges Racinet authored
This method is mostly used to check if the repository is emtpy (and that is in the case of Mercurial logically equivalent), but it's better to implement it using the GitLab branches file as it avoids actually opening the repository, spawning a thread for blocking iteration etc.
-
- Jul 29, 2023
-
-
Georges Racinet authored
This finally puts together all the common logic already used in `GetBlob`, and we can switch on the Comparison tests (one single test function for both gRPC methods). Using `blocking_gitlab_revision_node_prefix` is not very elegant, we could probably resolve all the revisions before entering the blocking thread (and perhaps also avoid scanning the state files many times), but this is deemed not so important for the time being, compared to the huge efficiency boost provided to Heptapod by implementing this method in Rust. As is now customary, the tracing part has us create a wrapper struct and implement `Debug` because of the bytes in `RevisionPath`. At some point we'll have to come up with something more generic for this.
-
Georges Racinet authored
This will be used once in a thread created by `spawn_blocking`, typically with the closure given to the likes of `repository::load_repo_and_stream`. It is a bit annoying to introduce the dependency to `futures::executor` for this, but we don't feel at ease using `tokio::runtime::Runtime.block_on`, as this is really meant as an entry point (looks like we'd need to create a new runtime and it's not clear whether it would interfere with the current running one or not).
-
Georges Racinet authored
This builds on the previously introduced repository file extraction and conversion to `GetBlobResponse`. It is important to emit at least one response, even if the resulting data is empty (usually because of `limit=0`).
-
Georges Racinet authored
This handles the fact that `GetBlobResponse` and `GetBlobsResponse` have some similar logic (including metadata only on the first message for a given blob) and some differences (more metadata in the latter). We enclose the common logic in the new `message::BlobResponseChunk` trait, and implement it for both response types. The intermixing with data is not very natural in the Rust ownership context, we try and make it not too painful. There is possibly room for improvement in reducing cloning/allocations, perhaps using the `Bytes`-oriented variant of `prost`.
-
Georges Racinet authored
This provides the base logic to access file data, and a common representation for metadata that will be enough to implement the `GetBlob` and `GetBlobs` gRPC request, and that we hope to be reusable for the likes of `TreeEntry` and `GetTreeEntries`. The `rhgitaly::message` module will provide the necessary conversions to actual Gitaly response types.
-
Georges Racinet authored
Same as in `hgitaly.streaming`, and ultimately as in Gitaly, this is used as a constant that can be overridden by environment variable.
-
Georges Racinet authored
-
- Jul 28, 2023
-
-
Georges Racinet authored
Same purpose as Python `hgitaly.oid`, of course more strongly typed.
-
- Jul 25, 2023
-
-
Georges Racinet authored
This is the removal of previously unneeded blocks kept for indentation.
-
Georges Racinet authored
Reduced duplication, and simpler error treatment.
-
Georges Racinet authored
Again keeping an unneeded block to get a readable diff.
-
Georges Racinet authored
This function provides the path to the store vfs, taking care of performing all checks, in particular that the repository exists. One advantage is the collapsing of the various problems (missing repo specfication, repo not found...) into a single `RepoSpecError`, so that callers can simply use `map_err()` and we don't need to take an error treatment argument. The check for `None` spares us a conditional block in the service implementation (kept an artificial block to have a readable diff by avoiding reindent). This is a new, simpler pattern, we'll see to generalize it.
-
- Jul 29, 2023
-
-
Georges Racinet authored
This is more convenient than using grpcurl to check the version. Also, one day it will be displayed in Heptapod UI, but that is not done yet.
-
- Jul 05, 2023
-
-
Georges Racinet authored
Closes #141
-