transaction: do not rely on a global variable to post_finalize file We can just add a new argument to the `addfilegenerator` function. This is more explicit and therefor clearer and less error prone. Differential Revision: https://phab.mercurial-scm.org/D12125

dirstate: introduce a "tracked-key" feature A new format variant is introduced. When used, a `tracked-key` file will be generated. That file will be update when the set of tracked file might have changed. This will be useful for external automation (e.g. build tool) to be notified when the set of relevant files changes. One of the motivation for this changes is to mitigate effect dirstate-v2 has on such automation. Since the dirstate file is updated much more frequently on dirstate-v2, monitoring update to that file is no longer a viable strategy. See the associated documentation for details about the feature To prevent older client to update the repository without updating that file, a new requirements is introduced. The `postfinalizegenerators` business is a bit weird, so I'll likely clean that up soon. Differential Revision: https://phab.mercurial-scm.org/D12124

dirstate: rename the filegenerator used for writing We will need a different name in the next changesets. Changing the name is actually not that trivial so we do it in its own changeset. Differential Revision: https://phab.mercurial-scm.org/D12123

dirstate: use a context manager when writing the dirstate This make sure the file is closed in a timely manner. We define a lambda for the file opening. It might seems a bit overkill here but a future changeset will need to do more of those. Differential Revision: https://phab.mercurial-scm.org/D12122

branching: merge stable into default

Added signature for changeset 75676122c2bf

Added tag 6.0.2 for changeset 75676122c2bf

relnotes: add 6.0.1 and 6.0.2 notes Differential Revision: https://phab.mercurial-scm.org/D12130

branchmap: stop writing cache for uncommitted data If we are about to write the branch while a transaction is active. we delay that write. After the transaction is closed, we flush all the write we delayed (unless they have been written in between). Differential Revision: https://phab.mercurial-scm.org/D12128

transaction: add a way to know a transaction has been finalized This will be useful to fix the timing of the branchmap on disk caching. Differential Revision: https://phab.mercurial-scm.org/D12127

branchmap: Add a test about writing branchmap and aborted transaction It turn out that we update the branchmap on disk whenever we recompute it… including in a middle of a transaction. This means writing the new branchmap too early (the changeset are not visible yet) and worse… it mean writing branchmap for transaction we rollback. so we introduce a test to highlight the issue (and prevent it to regress in the future). Differential Revision: https://phab.mercurial-scm.org/D12126

dirstate-v2: rename the configuration to enable the format The rename of the old experimental name was overlooked before the 6.0 release. We rename everything to use the new name (and keep the released name as an alias for compatibility). Differential Revision: https://phab.mercurial-scm.org/D12129

arbitraryfilectx: use our existing helpers for reading and writing files Differential Revision: https://phab.mercurial-scm.org/D12090

fix: remove unnecessary and overly strict check for divergence `rewriteutil.precheck()` checks for divergence these days, so we can remove the redundant check in `hg fix`. Differential Revision: https://phab.mercurial-scm.org/D12088

encoding: fix trim() to be O(n) instead of O(n^2) `encoding.trim()` iterated over the possible lengths smaller than the input and created a slice for each. It then calculated the column width of the result, which is of course O(n), so the overall algorithm was O(n). This patch rewrites it to iterate over the unicode characters, keeping track of the length so far. Also, the old algorithm started from the end of the string, which made it much worse when the input is large and the limit is small (such as the typical 72 we pass to it). You can time it by running something like this: ``` time python3 -c 'from mercurial.utils import stringutil; print(stringutil.ellipsis(b"0123456789" * 1000, 5))' ``` That drops from 4.05 s to 83 ms with this patch (and most of that is of course startup time). Differential Revision: https://phab.mercurial-scm.org/D12089

stream-clone: filter possible missing requirements using all supported one The `supportedformat` requirements is missing some important requirements and it seems better to filter out with all requirements we know, not just an "arbitrary" subset. The `supportedformat` set is lacking some important requirements (for example `revlog-compression-zstd`). This is getting fixed on default (for Mercurial 6.1) However, fixing that in 6.1 means the stream requirements sent over the wire will contains more items. And if we don't apply this fix on older version, they might end up complaining about lacking support for feature they actually support for years. This patch does not fix the deeper problem (advertised stream requirement lacking some of them), but focus on the trivial part : Lets use the full set of supported requirement for looking for unsupported ones. This patch should be simple to backport to older version of Mercurial and packager should be encouraged to do so. This is a graft of d9017df70135 from default. Differential Revision: https://phab.mercurial-scm.org/D12091

share-safe: enable by default (BC) The feature have been around for a year (4 version) and is quite important. Lets make it enabled by default. Differential Revision: https://phab.mercurial-scm.org/D11997

share-safe: add support for static-http repository We need to read the second requirements file in the static-http case too. Otherwise, static-http would miss most of the requirements and misbehave. Differential Revision: https://phab.mercurial-scm.org/D11996

legacy-revlog: fix requirement computation when cloning legacy repo The oldest format of repository does not have requirements so we need to treat them differently when cloning such repository. The previous code had issue whenever we start using a working-copy-only requirements. The "legacy" format is signaled using an empty requirements list. If we add working-copy-only requirements to it, this is no longer empty, and no longer legacy. The new code fix this, and will get fully tested in a couple of changeset, once the share-safe become the default. Differential Revision: https://phab.mercurial-scm.org/D11995

requirements: do not warn about dropping share-safe, unless explicitly set If we are just altering the default value, this is "fine". This will get tested once share-safe become the default. Differential Revision: https://phab.mercurial-scm.org/D11994

sparse: take lock before writing requirements with `share-safe`, we will also write file in the store. So we take the `lock` before altering the requirements. Differential Revision: https://phab.mercurial-scm.org/D11993

largefiles: take lock before writing requirements With `share-safe`, we will also write file in the store. We now take the `lock` as needed. Differential Revision: https://phab.mercurial-scm.org/D11992

lfs: take lock before writing requirements With `share-safe`, we will also write file in the store, so we take the store lock before writing requirements. Differential Revision: https://phab.mercurial-scm.org/D11991

test: simpler requirements matching in test-repo-compengines.t We focus on the requirements relevant for this tests. Differential Revision: https://phab.mercurial-scm.org/D11990

test: simpler requirements matching in test-lfs-largefiles.t We focus on the requirements relevant for this tests. Differential Revision: https://phab.mercurial-scm.org/D11989

test: simpler requirements matching in test-sidedata.t We focus on the requirements relevant for this tests. Differential Revision: https://phab.mercurial-scm.org/D11988

test: simpler requirement matching in test-revlog-v2.t We narrow the matching toward the one we actually care about. Differential Revision: https://phab.mercurial-scm.org/D11987

test-copies: narrow the format check in test-copies-in-changeset.t This will reduce the noise when touching unrelated format variants. Differential Revision: https://phab.mercurial-scm.org/D11986

tests: use debugrequires instead of grepping the file directly With `share-safe`, the requirements are stored in multiple files so it seems better to use the command that retrieve the information wherever it is. Differential Revision: https://phab.mercurial-scm.org/D11985

tests: use debugrequires in test-upgrade-repo.t Whenever possible, we use the `hg debugrequires` command instead of directly grepping the file. Differential Revision: https://phab.mercurial-scm.org/D11984

test: do not use `which` to find an exec path in the test Posix is deprecated `which` in favor of `command -v`… which does not provide the same feature. Debian is warning about this deprecation and this get annoying. Differential Revision: https://phab.mercurial-scm.org/D11983

stream-clone: stop considering working copy only requirements Now that the ducks are properly aligned on there shelve, it becomes easy to fix the list of requirements to consider for stream. We remove the working copy specific ones and add some missing revlog specific ones. Differential Revision: https://phab.mercurial-scm.org/D11982

stream-clone: also filter the requirement we put in the bundle 2 We were wrongly putting irrelevant requirements in the bundle and the receiving side was getting confused, treating them as being missing while still putting them in the `requires` file. Leading do corrupted repositories. This changes fix stream-clone behavior regarding format when bundle-2 is involved, so we now also test this cases. Behavior with older version of Mercurial will be fine as they filter the requirements they get from the bundle on their side anyway. Differential Revision: https://phab.mercurial-scm.org/D12084

requirements: add an official `REVLOG_COMPRESSION_ZSTD` const Such constant was missing and its value was missing from the set of requirements that needs to be preserved through stream clone. This did not had any consequence yet as the "bundle 2 does not filter its requirements" is shadowing the issue. However we are now in a situation where we can fix this issue. So lets do it next. With the preparation work on test, changing the streamreqs value only impact two tests, where checking the full value seems to remains relevant. Important note: Since older version of Mercurial used the old `supportedformat` class attribute to check for stream requirement they supported, older version will consider this requirements to prevent them from using streaming clone. Even as they support this requirements for years. Pack for stable will be send to fix it, but they will have to be backported to older version if needed. Differential Revision: https://phab.mercurial-scm.org/D12083

stream-requirements: smoother matching in test-http-protocol.t The value of `streamreqs` changes depending of the format used, but this is not relevant for this tests. So we doing a smarter matching of the line to avoid this test interfering with format changes in the future. Differential Revision: https://phab.mercurial-scm.org/D12082

stream-requirements: smoother matching in test-ssh-bundle1.t The value of `streamreqs` changes depending of the format used, but this is not relevant for this tests. So we doing a smarter matching of the line to avoid this test interfering with format changes in the future. Differential Revision: https://phab.mercurial-scm.org/D12081

stream-requirements: smoother matching in test-ssh-proto.t The value of `streamreqs` changes depending of the format used, but this is not relevant for this tests. So we doing a smarter matching of the line to avoid this test interfering with format changes in the future. Differential Revision: https://phab.mercurial-scm.org/D12080

stream-requirements: smoother matching in test-hgweb-commands.t The value of `streamreqs` changes depending of the format used, but this is not relevant for this tests. So we doing a smarter matching of the line to avoid this test interfering with format changes in the future. Differential Revision: https://phab.mercurial-scm.org/D12079

stream-requirements: smoother matching in test-treemanifest.t The value of `streamreqs` changes depending of the format used, but this is not relevant for this tests. So we doing a smarter matching of the line to avoid this test interfering with format changes in the future. Differential Revision: https://phab.mercurial-scm.org/D12078

stream-requirements: smoother matching in test-ssh-proto-unbundle.t The value of `streamreqs` changes depending of the format used, but this is not relevant for this tests. So we doing a smarter matching of the line to avoid this test interfering with format changes in the future. Differential Revision: https://phab.mercurial-scm.org/D12077

test-bundle: split some variant in there own section Same rational as the previous commit about test-bundle.t. These line are quite volatile and having dedicated block will make their update clearer and simpler. Differential Revision: https://phab.mercurial-scm.org/D12076

test-bundle: split each variant in there own section Having conditional line matching for this kind of big block is quite inconvenient to update. The new way should be simpler to maintain. Differential Revision: https://phab.mercurial-scm.org/D12075

test-http-bad-server: abstract the `streamreqs` value Now that socket closing are decided much closer to the actual closure, we can use fuzzy matching for this part of the exchange that changes on a regular basis. This should make fixing bug in stream requirements and enabling new format much simpler. Differential Revision: https://phab.mercurial-scm.org/D12074

test-http-bad-server: use the new pattern-reading for a test-case This test case is now less sensitive to change of unrelated bits of the client/server exchange. Since this introduce some churn in the output, we do it independently for each test cases. This patch is the last of such changes, for both sent and recv cases. Differential Revision: https://phab.mercurial-scm.org/D12073

test-http-bad-server: use the new pattern-reading for a test-case This test case is now less sensitive to change of unrelated bits of the client/server exchange. Since this introduce some churn in the output, we do it independently for each test cases. Differential Revision: https://phab.mercurial-scm.org/D12072

test-http-bad-server: use the new pattern-reading for a test-case This test case is now less sensitive to change of unrelated bits of the client/server exchange. Since this introduce some churn in the output, we do it independently for each test cases. Differential Revision: https://phab.mercurial-scm.org/D12071

test-http-bad-server: use the new pattern-reading for a test-case This test case is now less sensitive to change of unrelated bits of the client/server exchange. Since this introduce some churn in the output, we do it independently for each test cases. Differential Revision: https://phab.mercurial-scm.org/D12070

test-http-bad-server: use the new pattern-reading for a test-case This test case is now less sensitive to change of unrelated bits of the client/server exchange. Since this introduce some churn in the output, we do it independently for each test cases. Differential Revision: https://phab.mercurial-scm.org/D12069

test-http-bad-server: introduce socket closing after reading a pattern We introduce the `close-after-recv-patterns` option. It has the same goal as `close-after-send-patterns` with a slightly different implementation. Reading "up to a pattern" is hard. As we can only check the pattern from what we have already read (inlike writing, were we can check what we are about to write). So instead we make the `close-after-recv-patterns` alter the behavior of the existing `close-after-recv-bytes`. The value from `close-after-recv-bytes` only gets into play after we have seen the pattern from `close-after-recv-patterns`. This allow us to achieve the target benefit without changing the read pattern too much. Differential Revision: https://phab.mercurial-scm.org/D12068

test-http-bad-server: move on test from bytes-count to pattern This specific test case is now using the new "close when a pattern is seen" approach compared to the more fragile "that many bytes were sent" approach. Since such change are still a bit noisy, we split each change in there own changesets. Differential Revision: https://phab.mercurial-scm.org/D12067

test-http-bad-server: move on test from bytes-count to pattern This specific test case is now using the new "close when a pattern is seen" approach compared to the more fragile "that many bytes were sent" approach. Since such change are still a bit noisy, we split each change in there own changesets. Differential Revision: https://phab.mercurial-scm.org/D12066

test-http-bad-server: move on test from bytes-count to pattern This specific test case is now using the new "close when a pattern is seen" approach compared to the more fragile "that many bytes were sent" approach. Since such change are still a bit noisy, we split each change in there own changesets. Differential Revision: https://phab.mercurial-scm.org/D12065

test-http-bad-server: move on test from bytes-count to pattern This specific test case is now using the new "close when a pattern is seen" approach compared to the more fragile "that many bytes were sent" approach. Since such change are still a bit noisy, we split each change in there own changesets. Differential Revision: https://phab.mercurial-scm.org/D12064

test-http-bad-server: move on test from bytes-count to pattern This specific test case is now using the new "close when a pattern is seen" approach compared to the more fragile "that many bytes were sent" approach. Since such change are still a bit noisy, we split each change in there own changesets. Differential Revision: https://phab.mercurial-scm.org/D12063

test-http-bad-server: move on test from bytes-count to pattern This specific test case is now using the new "close when a pattern is seen" approach compared to the more fragile "that many bytes were sent" approach. Since such change are still a bit noisy, we split each change in there own changesets. Differential Revision: https://phab.mercurial-scm.org/D12062

test-http-bad-server: move on test from bytes-count to pattern This specific test case is now using the new "close when a pattern is seen" approach compared to the more fragile "that many bytes were sent" approach. Since such change are still a bit noisy, we split each change in there own changesets. Differential Revision: https://phab.mercurial-scm.org/D12061

test-http-bad-server: move on test from bytes-count to pattern This specific test case is now using the new "close when a pattern is seen" approach compared to the more fragile "that many bytes were sent" approach. Since such change are still a bit noisy, we split each change in there own changesets. Differential Revision: https://phab.mercurial-scm.org/D12060

test-http-bad-server: move on test from bytes-count to pattern This specific test case is now using the new "close when a pattern is seen" approach compared to the more fragile "that many bytes were sent" approach. Since such change are still a bit noisy, we split each change in there own changesets. Differential Revision: https://phab.mercurial-scm.org/D12059

test-http-bad-server: move on test from bytes-count to pattern This specific test case is now using the new "close when a pattern is seen" approach compared to the more fragile "that many bytes were sent" approach. Since such change are still a bit noisy, we split each change in there own changesets. Differential Revision: https://phab.mercurial-scm.org/D12058

test-http-bad-server: move on test from bytes-count to pattern This specific test case is now using the new "close when a pattern is seen" approach compared to the more fragile "that many bytes were sent" approach. Since such change are still a bit noisy, we split each change in there own changesets. Differential Revision: https://phab.mercurial-scm.org/D12057

test-http-bad-server: move on test from bytes-count to pattern This specific test case is now using the new "close when a pattern is seen" approach compared to the more fragile "that many bytes were sent" approach. Since such change are still a bit noisy, we split each change in there own changesets. Differential Revision: https://phab.mercurial-scm.org/D12056

test-http-bad-server: move on test from bytes-count to pattern This specific test case is now using the new "close when a pattern is seen" approach compared to the more fragile "that many bytes were sent" approach. Since such change are still a bit noisy, we split each change in there own changesets. Differential Revision: https://phab.mercurial-scm.org/D12055

test-http-bad-server: modify some `tail` call fit their goal We want to check that we stop writing when we want to stop writing. Differential Revision: https://phab.mercurial-scm.org/D12054

test-http-bad-server: move on test from bytes-count to pattern This specific test case is now using the new "close when a pattern is seen" approach compared to the more fragile "that many bytes were sent" approach. Since such change are still a bit noisy, we split each change in there own changesets. Differential Revision: https://phab.mercurial-scm.org/D12053

test-http-bad-server: move on test from bytes-count to pattern This specific test case is now using the new "close when a pattern is seen" approach compared to the more fragile "that many bytes were sent" approach. Since such change are still a bit noisy, we split each change in there own changesets. Differential Revision: https://phab.mercurial-scm.org/D12052

test-http-bad-server: move on test from bytes-count to pattern This specific test case is now using the new "close when a pattern is seen" approach compared to the more fragile "that many bytes were sent" approach. Since such change are still a bit noisy, we split each change in there own changesets. Differential Revision: https://phab.mercurial-scm.org/D12051

test-http-bad-server: move on test from bytes-count to pattern This specific test case is now using the new "close when a pattern is seen" approach compared to the more fragile "that many bytes were sent" approach. Since such change are still a bit noisy, we split each change in there own changesets. Differential Revision: https://phab.mercurial-scm.org/D12050

test-bad-http-server: introduce sock closing when writing a pattern Instead of using a fixed number a bytes, a very unstable method. We introduce a new config that can define a regex. One this regex is detected, nothing after it will be written and the socket will be closed. Tests will be migrated to this new method in later changesets. Differential Revision: https://phab.mercurial-scm.org/D12049

test-http-bad-server: document that the value are actually a list I don't really expect this doc to be read a lot, but mentioning that this is a list seems important. Differential Revision: https://phab.mercurial-scm.org/D12048

test-http-bad-server: refactor the writing logic to avoid early return Our ultimate goal is to add another way to define the connection needs to be closed. To do so, we need the "read" code to be more unified. Differential Revision: https://phab.mercurial-scm.org/D12047

test-http-bad-server: replace the default 65537 value in output This is the default block value that we use here and it make the output overall less readable. So we replace it with `~` Differential Revision: https://phab.mercurial-scm.org/D12046

test-http-bad-server: refactor the reading logic to avoid early return Our ultimate goal is to add another way to define the connection needs to be closed. To do so, we need the "read" code to be more unified. Differential Revision: https://phab.mercurial-scm.org/D12045

test-http-bad-server: factor code dealing with "read" in the new object This will make sure both `read` and `readline` do the same processing and make it simpler to update that processing in the future. Differential Revision: https://phab.mercurial-scm.org/D12044

test-http-bad-server: factor code dealing with "write" in the new object This will make sure both `sendall` and `write` do the same processing and make it simpler to update that processing in the future. Differential Revision: https://phab.mercurial-scm.org/D12043

test-http-bad-server: unify log printing for `sendall` and `write` The `write` function was logging before calling the function while the `sendall` function were logging after calling the function. We align the `write` behavior on the `sendall` behavior before unifying the code. Differential Revision: https://phab.mercurial-scm.org/D12042

test-http-bad-server: track close condition in an object In order to make the logic more advanced, we need to unify it. To unify it, we introduce a small object that will be responsible for tracking and enforcing "premature socket close" conditions for both proxy object (socketproxy and fileobjectproxy). More logic will be moved into the object in later changesets. Differential Revision: https://phab.mercurial-scm.org/D12041

test-http-bad-server: use more readable name for variable This make the code more accessible overall. Differential Revision: https://phab.mercurial-scm.org/D12040

test-http-bad-server: align output by using `;` This is the only message using a `,` instead of `;` and that seems unintentional. I am unifying the message before unigying the code. Differential Revision: https://phab.mercurial-scm.org/D12039

test-http-bad-server: highlight each test with section Small "documentation" change to highlight each individual tests better. Differential Revision: https://phab.mercurial-scm.org/D12038

test-http-bad-server: rename config to use `-` This make things much more readable. Differential Revision: https://phab.mercurial-scm.org/D12037

test-http-bad-server: move the extension in `testlib` This seems like a better location for it. Differential Revision: https://phab.mercurial-scm.org/D12036

requirements: remove the `localrepo.supportedformat` attribute It is not longer used anywhere. Differential Revision: https://phab.mercurial-scm.org/D12035

requirements: sort _basesupported This will make the next changeset clearer. Differential Revision: https://phab.mercurial-scm.org/D12034

requirements: move the comment about manifestv2 in the module This is about requirements, so lets put it into the requirements module. We are about to remove code it was previously attached too. Differential Revision: https://phab.mercurial-scm.org/D12033

stream-clone: add a explicit set list requirements relevant to stream clone This set explicitly lists all the requirements that "stream clone" will preserve. Any other one will not be preserved. The role of listing the relevant one was previously filled by `repo.supportedformat`, but it seems clearer to use that such global and explicit set. The `repo.supportedformat` attribute will be cleaned up in a later changeset The true meaning of `repo.supportedformat` vs `repo._basesupported` was lost over time so, the content is currently bad. For example, `dirstate-v2` is currently considered relevant to the stream clone, or internal phase is missing. We kept the same content in this changeset and we will fix them later. Differential Revision: https://phab.mercurial-scm.org/D12032

stream-clone: filter possible missing requirements using all supported one The `supportedformat` requirements is missing some important requirements and it seems better to filter out with all requirements we know, not just an "arbitrary" subset. This is especially relevant as we about to phase out the `supportedformat` class attribute. (a backport of this change to stable should be sent soon). Differential Revision: https://phab.mercurial-scm.org/D12031

requirements: move "bookmark in store" requirements in the right module There seems to be no reasons to not have it with the others. Differential Revision: https://phab.mercurial-scm.org/D12030

stream-clone: add a explicit test for format change during stream clone They are different kind of requirements, the one which impact the data storage and are relevant to the files being streamed and the one which does not. For example some requirements are only relevant to the working copy, like sparse, or dirstate-v2. Since they are irrelevant to the content being streamed, they do not prevent the receiving side to use streaming clone and mercurial skip adverting them over the wire and, ideally, within the bundle. In addition, this let the client decide to use whichever format it desire for the part that does not affect the store itself. So the configuration related to these format are used as normal when doing a streaming clone. In practice, the feature was not really tested and is badly broken with bundle-2, since the requirements are not filtered out from the stream bundle. So we start with adding simple tests as a good base before the fix and adjust the feature. Differential Revision: https://phab.mercurial-scm.org/D12029

stream-clone: factor computation of requirement of a stream clone This gather code duplicated in multiple place and will make it easier to modify it safely in the future. Differential Revision: https://phab.mercurial-scm.org/D12028

stream-clone: factor computation of new clone requirement out This gather code duplicated in multiple place and will make it easier to modify it safely in the future. Differential Revision: https://phab.mercurial-scm.org/D12027

stream-clone: drop some outdated lines in test-stream-bundle-v2.t These line refer to a "dirstate-v2" variant that is no longer in use. The associated line drifted out of sync with reality. Differential Revision: https://phab.mercurial-scm.org/D12026

relnotes: add note about changed labels texts Differential Revision: https://phab.mercurial-scm.org/D12085

graft: attempt to make merge labels more helpful Differential Revision: https://phab.mercurial-scm.org/D12025

shelve: attempt to make merge labels more helpful Differential Revision: https://phab.mercurial-scm.org/D12024

histedit: attempt to make merge labels more helpful Differential Revision: https://phab.mercurial-scm.org/D12023

update: set custom conflict label for base commit Differential Revision: https://phab.mercurial-scm.org/D12022