share: rework config options to be much clearer and easier Recently I implemented various boolean configs which control how to behave when there is a share-safe mismatch between source and share repository. Mismatch means that source supports share-safe where as share does not or vice versa. However, while discussion and documentation we realized that it's too complicated and there are some combinations of values which makes no sense. We decided to introduce a config option with 4 possible values which makes controlling and understanding things easier. The config option `share.safe-mismatch.source-{not-}safe` can have following 4 values: * abort (default): error out if there is mismatch * allow: allow to work with respecting share source configuration * {up|down}grade-abort: try to {up|down}grade, if it fails, abort * {up|down}grade-allow: try to {up|down}grade, if it fails, continue in allow mode I am not sure if I can explain 3 config options which I deleted right now in just 5 lines which is a sign of how complex they became. No test changes demonstrate that functionality is same, only names have changed. Differential Revision: https://phab.mercurial-scm.org/D9785

share: move share safe functionality out of experimental The share-safe functionality is complete and all configuration options are implemented. The behavior is well discussed on mailing list and in reviews. Let's unmark this as experimental to solve a chichen and egg issue. Differential Revision: https://phab.mercurial-scm.org/D9823

share: rename share-safe warning config Config introduced in previous patch was `share.source-safe-mismatch`. Let's rename the warn as `share.source-safe-mismatch.warn`. While we are here, made sure we have different configs for upgrade and downgrade. Differential Revision: https://phab.mercurial-scm.org/D9786

share: rework config options to be much clearer and easier Recently I implemented various boolean configs which control how to behave when there is a share-safe mismatch between source and share repository. Mismatch means that source supports share-safe where as share does not or vice versa. However, while discussion and documentation we realized that it's too complicated and there are some combinations of values which makes no sense. We decided to introduce a config option with 4 possible values which makes controlling and understanding things easier. The config option `share.safe-mismatch.source-{not-}safe` can have following 4 values: * abort (default): error out if there is mismatch * allow: allow to work with respecting share source configuration * {up|down}grade-abort: try to {up|down}grade, if it fails, abort * {up|down}grade-allow: try to {up|down}grade, if it fails, continue in allow mode I am not sure if I can explain 3 config options which I deleted right now in just 5 lines which is a sign of how complex they became. No test changes demonstrate that functionality is same, only names have changed. Differential Revision: https://phab.mercurial-scm.org/D9785

localrepo: disallow share if there is a version mismatch by default Earlier we used to allow shares which don't use share-safe mechanism to access repository which uses share-safe mechanism. This defeats the purpose and is bad behavior. This patch disallows that. Next patch will introduce a config option to allow that and have clearer understanding around various options. Differential Revision: https://phab.mercurial-scm.org/D9784

upgrade: re-read current requirements after taking lock Since we are writing to repository, it's better to re-read after taking the lock. Differential Revision: https://phab.mercurial-scm.org/D9822

upgrade: take lock only for part where it's required The final config calculation code does not require a lock, only writing it back does require one. Differential Revision: https://phab.mercurial-scm.org/D9783

clang-format: reorder includes to appease the formatter The bad order was introduced in d0225a22040c. Differential Revision: https://phab.mercurial-scm.org/D9829

run-tests: catch a Windows specific error when testing for a free socket I'm not sure why this only happens with py3, but this error code doesn't map to any of the 3 currently being handled, and kills `run-tests.py` before it can run any tests when it happens: OSError: [WinError 10013] An attempt was made to access a socket in a way forbidden by its access permissions The documentation[1] says this can happen if another process is bound to the address with exclusive access. This seems to keep it happy. [1] https://docs.microsoft.com/en-us/windows/win32/winsock/windows-sockets-error-codes-2 Differential Revision: https://phab.mercurial-scm.org/D9816

run-tests: work around the Windows firewall popup for server processes Windows doesn't have a `python3` executable, so cc0b332ab9fc attempted to work around the issue by copying the current python to `python3.exe`. That put it in `_tmpbindir` because of failures in `test-run-tests.t` when using `_bindir`, which looked like a process was trying to open it to write out a copy while it was in use. (Interestingly, I couldn't reproduce this running the test by itself in a loop for a couple of hours, but it happens constantly when running all tests.) The problem with using `_tmpbindir` is that it is the randomly generated path for the test run, and instead of Windows Firewall remembering the executable signature or image hash when allowing the process to open a server port, it apparently remembers the image path. That means every run will trigger a popup to allow it, which is bad for firing off a test run and walking away. I tried to symlink to the python executable, but that currently requires admin priviledges[1]. This will prompt the first time if the underlying python binary has never opened a server port, but appears to avoid it on subsequent runs. [1] https://bugs.python.org/issue40687 Differential Revision: https://phab.mercurial-scm.org/D9815

hghave: split apart testing for the curses module and `tic` executable ef771d329961 skipped the check for the `tic` executable, because the curses module alone on Windows is enough to pass the `test-*-curses.t` tests. However, `test-status-color.t` uses this same check and explicitly invoked the executable, which fails on Windows. From the cursory searching I did, curses on unix requires `tic`, which I assume is why they were tied together in the first place. So this continues to require both to get past the curses guards on non Windows platforms. Differential Revision: https://phab.mercurial-scm.org/D9814

sqlitestore: disable test with chg There are known issues with transactions not being closed in a timely fashion, making the test flakey. Differential Revision: https://phab.mercurial-scm.org/D9821

heptapod-ci: allow testing with docker image other than :latest The project have a default config of latest for this variable. Differential Revision: https://phab.mercurial-scm.org/D9792

persistent-nodemap: also exchange the nodemap data over the wire It appears that 2 bytes are never equals to 3 bytes. Differential Revision: https://phab.mercurial-scm.org/D9810

persistent-nodemap: catch the right exception on python On Python 2, Missing file are IOError, not OSError, apparently. Differential Revision: https://phab.mercurial-scm.org/D9809

hghave: adjust the definition of `tic` to allow curses tests on Windows Might as well pick up the test coverage if it works as-is. Differential Revision: https://phab.mercurial-scm.org/D9808

windows: wrap `os.getcwd()` in `os.path.realpath()` on py3 I noticed various `test-check-*` failures that were printing absolute paths when repo relative paths were expected. This was due to the drive letter in `repo.root` being uppercased as it is run through `os.path.realpath()`, and then the simple string comparison against the (lowercased) `_cwd` member of dirstate in `dirstate.getcwd()` causing an absolute path to be returned, instead of the expected `b''`. That in turn causes `scmutil.getuipathfn()` to wrongly use `repo.pathto()` with an absolute cwd path. . Differential Revision: https://phab.mercurial-scm.org/D9806

packaging: include `windows-curses` on Windows Otherwise the interface degrades to the old hunk selection method on Windows when `ui.tweakdefaults=1`. I've been using this for a couple of months without any issues, other than it seems to toggle off edit mode for the terminal when run through MSYS. Using cmd.exe seems to work fine. Differential Revision: https://phab.mercurial-scm.org/D9805

cext: fix compiler error in revlog.c on Windows with py2 Visual Studio 2008 doesn't have <stdint.h>, and we worked around it before with this header. Differential Revision: https://phab.mercurial-scm.org/D9804

discovery: add a devel.discovery.exchange-heads Currently all discovery start with testing local and remote heads. For analysis purpose we make it possible to disable that initial "handshake" and start discovery with the whole repository as undecided. Differential Revision: https://phab.mercurial-scm.org/D9801

test-regression: backing out D9640 (63c923fd7fa8) Since D9640 was pushed, we have seen failure on at least: - test-commit-interactive.t - test-i18n.t - test-log.t The author that change requested it to be removed until he can investigate. For reference: https://phab.mercurial-scm.org/D9640 Backed out changeset 63c923fd7fa8 Differential Revision: https://phab.mercurial-scm.org/D9803

clone: make sure we warm the cache after a clone This work around any deviciency/limitation of the clone process. In our case this ensure the persistent nodemap exist with valid content. Ideally, the cloning process would also do "the right thing". However since older server will never be able to do "the right thing". The local workaround will be necessary anyway. I am not worried by the performance impact of this as `hg clone` is non-instant on large repositories where is could matters. Warming the cache if they are already correct is very fast. And if they are not already warm, this seems like a good time to do so. This impact various test as more cache are now warmed sooner, all the change should be harmless. Differential Revision: https://phab.mercurial-scm.org/D9789

persistent-nodemap: remove the "experimental" tag to the feature This has been around for a while and has all the discussed configuration options. Differential Revision: https://phab.mercurial-scm.org/D9764

persistent-nodemap: drop the storage.revlog.nodemap.mode config We now have a new, better, higher level, option. So we can drop the older one as planned. Differential Revision: https://phab.mercurial-scm.org/D9763

persistent-nodemap: add a "abort" option to the slow-path config We make it the default, and document the behavior in the help for the main config option. Differential Revision: https://phab.mercurial-scm.org/D9762

persistent-nodemap: add a "warn" option to the slow-path config And make it the default until we get an abort option. Differential Revision: https://phab.mercurial-scm.org/D9761

persistent-node: check the value of the slow-path config We should probably provide some standard for this in config item, but this is a quest for another adventure. Differential Revision: https://phab.mercurial-scm.org/D9760

persistent-nodemap: add a revlog.storage.persistent-nodemap.slow-path option As discussed during the sprint, we want to prevent user to get an unexpected performance regression when accessing a repository using "persistent-nodemap" without the associated Rust extension. We start by adding a config declaration and some documentation. Since "allow" is the current behavior, we don't need to add any code. The option possible value will come later. Note that we already have a `storage.revlog.nodemap.mode` option, but that option is a bit different. It does some warning and checking at revlog instantiation time. While we want something done at requirements checking time. Since we plan for new names and new config value names, we introduce a new option and will drop the old one later. Differential Revision: https://phab.mercurial-scm.org/D9758

persistent-nodemap: document storage.revlog.persistent-nodemap.mmap=no The feature is gaining public facing document, its options too. Differential Revision: https://phab.mercurial-scm.org/D9757

persistent-nodemap: rename the storage.revlog.nodemap.mmap option Since the feature is publicly known as "persistent-nodemap" this seems better to reference it as "persistent-nodemap" in related config too. So we rename the option to `storage.revlog.persistent-nodemap.mmap`. Differential Revision: https://phab.mercurial-scm.org/D9756

persistent-nodemap: document the feature in `hg help config.format` This is necessary before we advertise the feature more. Differential Revision: https://phab.mercurial-scm.org/D9755

persistent-nodemap: write down the sprint conclusion This write down the plan for turning feature that are only fast when the rust extension exists. Future changesets will implement it. Differential Revision: https://phab.mercurial-scm.org/D9754