Added tag 6.5.3 for changeset c083d9776cb2

relnotes: add 6.5.3

revlog: avoid opening and closing the file for each cloned revision The previous code was flushing files after each new revision, slowing things down. For exemple, with this change, the evolve repository can run `hg debugupgraderepo --run --optimize re-delta-parent` in about 3.4s instead of 4.5 seconds.

censor: accept censored revision during upgrade They can simply be passed by as censored.

censor: show that censored revision prevent repository upgrade This is not great.

smartset: don't ignore hidden revs when intersecting This fixes the bug I demonstrated in the previous commit, but I'm not sure at all if it's the right way of doing it.

tests: demonstrate crash in `unstable()` with internal-phase orphans

rust-matchers: fix quadratic complexity in `FileMatcher` Concretely, this command: ``` $ echo hg up -r <nodeid>; time hg revert dir1 dir2 -r <othernode> --debug hg up -r <nodeid> real 0m14.690s user 0m14.766s sys 0m5.430s ``` was much slower despite using 16 cores before this change. The approach taken here is the same one used in match.py, in exactmatcher. This changeset was originally written by Valentin Gatien-Baron in a private repository. I have redacted the commit message and did a minor clean up of the code.

revlog: add a small cache of unfiltered chunk This can provides a massive boost to the reading of multiple revision and the computation of a valid delta chain. This greatly help operation like `hg log --patch`, delta computation (helping pull/unbundle), linkrev adjustment (helping copy tracing). A first round of benchmark for `hg log --patch --limit 1000` shows improvement in the 10-20% range on "small" repository like pypy or mercurial and large improvements (about 33%) for more complex ones like netbeans and mozilla's. These speeds up are consistent with the improvement to `hg pull` (from a server sending poor deltas) I saw benchmarking this last year. Further benchmark will be run during the freeze. I added some configuration in the experimental space to be able to further test the effect of various tuning for now. This feature should fit well in the "usage/resource profile" configuration that we should land next cycle. When it does not provides a benefit the overhead of the cache seem to be around 2%, a small price for the big improvement. In addition I believe we could shave most of this overhead with a more efficent lru implementation.

revlog: minor refactor in the chunk gather process We will introduce some caching in this method in the next changeset, we make some of the most "disruptive" change first as touching this could break (and maybe did during the development process).

changelog-delay: move the delay/divert logic inside the (inner) revlog Instead of hacking throught the vfs/opener, we implement the delay/divert logic inside the `_InnerRevlog` and `randomaccessfile` object. This will allow to an alternative implementation of the `_InnerRevlog` that does not need to use Python details. As a result, the new implementation can use the transaction less agressively and avoid some extra output since no data had been written yet. That seems like a good side effect.

revlog: add a `canonical_index_file` attribute on inner revlog This is currently the same and the index_file but it will become more complex when we start doing delayed write.

changelog-delay: move "delayed" check to a more official API To avoid reaching inside the inner object in the future, we needs some official API. We put one of such API early to reduce the size of the final diff.

changelog-delay: move the appender class next to randomaccessfile We want to move the delay/divert logic at the revlog level (to have all IO related logic in the _InnerRevlog) we start with small piece that are easy to move on their own.

changelog-delay: adds some check around delaying and diverting write Theses assert shows we never call delay or divert if the revlog hold file handle on the revlog's file.

revlog: consolidate cache invalidation within the inner objet The invalidation of data that belong to the inner object can and should be delegated to the inner object.

shelve: drop some weird manually stripping before transaction abort I cannot find a case where this is actually useful, so lets triggering strip in the middle of uncommitted transaction.

revlog: move entry writing in the inner object This is the second big piece of API we want to live on the inner object. With this the inner object have freedom it needs to implement things internally. (except for the delayed write feature of the changelog, that will be migrated in coming changesets)

revlog: move `sidedata` in the inner object Same object as rawtext, but for sidedata.

revlog: move the `rawtext` method on the inner object This is one of the current core goal of the inner object: having a "simple" API for some of the core revlog action, like getting a raw text, with the inner object able to do that part internally, on its own.

revlog: move the_revisioncache on the inner object The goal for this inner object is to compute that things we cache, it make more sense to have the inner object handle it directly.

revlog: move the `deltachain` method on the inner object This is a necessary step before being able to move more logic around restoring a revision content there. For now, we do a simple patch for the perf extension logic, when the implementation of the inner object changes, we will likely need some evolution of the API. However this is true of many things in the perf extension. So we will see this later.

revlog: move the `_chunks` method on the inner object This is a necessary step before being able to move more logic around restoring a revision content there. For now, we do a simple patch for the perf extension logic, when the implementation of the inner object changes, we will likely need some evolution of the API. However this is true of many things in the perf extension. So we will see this later.

revlog: add a couple more of useful method on the inner object This will be needed for the next changeset.

revlog: move the `_chunk` method on the inner object This is a necessary step before being able to move more logic around restoring a revision content there. For now, we do a simple patch for the perf extension logic, when the implementation of the inner object changes, we will likely need some evolution of the API. However this is true of many things in the perf extension. So we will see this later.

revlog: move the compression/decompression logic on the inner object This is a necessary step before being able to move more logic around restoring a revision content there. For now, we do a simple patch for the perf extension logic, when the implementation of the inner object changes, we will likely need some evolution of the API. However this is true of many things in the perf extension. So we will see this later.

revlog: move the splitting-inline-revlog logic inside the inner object This is another large IO block that we need to move within the inner object if we want's it to be self sufficient.

revlog: synchronise the various attribute holding the index filename The segmentfile and the attribute need to be synchronized, let's enforce that.

revlog: drop reference to docket in the inline-splitting code revlog with a docket do not use inline revlog and do not need to split them. So we can remove some code handling docket there.

revlog: move _getsegmentforrevs on the internal object See inline documentation for details.

revlog: create a iteration of a _InnerRevlog object within the revlog The goal of this object is to isolate a sub-API that can be implemented by a compiled object (e.g. Rust). So the boundary of this object will be arbitrary depending of what can we easily implemented in the Compiled code. For now, we start simple, and move the code that manage the IO objects in the inner object. More will come in the coming changesets. Note: the object definition could live in the different module to thin the `revlog.py` file, however there are other better candidate for extraction first and I have enought patch stacked on top of the this one for the split in this patch not to be worth it. So I leave this to future me.

revlog: drop the unused `_chunkcache` attribute Apparently, some time ago, the chunk cache moved in the randomaccessfile object.

revlog: drop the unused `_sidedatareadfp` method It has no caller anywhere and is probably the remains of some older code.

revlog: also migrates `revlog.upperboundcomp` to ConfigClass This was planned but overlooked when doing the rest of the migration.

revlog: small doc to the `files` method

revlog: remove the `_indexfp` method The method is only used two time and we want to move that code to a lower level object. So we simply inline the method to simplify further refactoring.

revlog: avoid opening and closing the file for each cloned revision The previous code was flushing files after each new revision, slowing things down. For exemple, with this change, the evolve repository can run `hg debugupgraderepo --run --optimize re-delta-parent` in about 3.4s instead of 4.5 seconds.

censor: accept censored revision during upgrade They can simply be passed by as censored.

censor: show that censored revision prevent repository upgrade This is not great.

stream-clone: fix a crash when a repo with an empty revlog is cloned

debug-delta-chain: print less data by default This is faster and simpler to read.

debugdeltachain: add a parameter to display all info This will be useful with the next changeset that change the defaul output to display the minimum amount of information.

debug-delta-chain: actually skip unrequested computation Doing quick test on mozilla-unified show we can run up to 2× faster by skipping some of these computation.

debug-delta-chain: add options to control what we compute Currently this mostly controls what we display, but actual computation saving will come soon.

debug-delta-chaing: add a parameter to select revision to look at This allows for much faster runtime when we are interrested in some revisions only.

delta-chain: extract some debugdeltachain logic is object Moving from a closure to an object's method will help us to extend the command logic.

delta-chain: move the debugdeltachain command in revlogutils There is a dedicated `mercurial.revlogutils.debug` module were this code fits well.

dirstate: document the `changing_*` context manager The methods that requires them have documentation, but the context themselves had none. This is now fixed.

revlog: deprecate the compatibility config property Now that core is no longer use them, we can deprecated them. Let us aim at removing them in the next version.

revlog: remove legacy usage of `_lazydeltabase` All core code is now getting the setting from the DeltaConfig object.

revlog: remove legacy usage of `_lazydelta` All core code is now getting the setting from the DeltaConfig object.

revlog: remove legacy usage of `canonical_parent_order` All core code is now getting the setting from the FeatureConfig object.

revlog: remove legacy usage of `_compute_rank` All core code is now getting the setting from the DeltaConfig object.

revlog: remove legacy usage of `_srmingapsize` All core code is now getting the setting from the DataConfig object.

revlog: remove legacy usage of `_srdensitythreshold` All core code is now getting the setting from the DataConfig object.

revlog: remove legacy usage of `hassidedata` All core code is now getting the setting from the FeatureConfig object.

revlog: remove legacy usage of `_sparserevlog` All core code is now getting the setting from the DeltaConfig object.

revlog: remove legacy usage of `_withsparseread` All core code is now getting the setting from the DeltaConfig object.

revlog: remove legacy usage of `_maxdeltachainspan` All core code is now getting the setting from the DeltaConfig object.

revlog: remove legacy usage of `_compengineopts` All core code is now getting the setting from the FeatureConfig object.