rust-chg: upgrade to 2018 edition and remove useless extern crates Differential Revision: https://phab.mercurial-scm.org/D8399

mergestate: store about files resolved in favour of other Committing a merge sometimes wrongly creates a new filenode where it can re-use an existing one. This happens because the commit code does it's own calculation and does not know what happened on merge. This starts storing information in mergestate about files which were automatically merged and the other/remote version of file was used. We need this information at commit to pick the filenode parent for the new commit. This issue was found by Pierre-Yves David and idea to store the relevant parts in mergestate is also suggested by him. Somethings which can be further investigated are: 1) refactoring of commit logic more to depend on this information 2) maybe a more generic solution? Differential Revision: https://phab.mercurial-scm.org/D8392

histedit: add test that an invalid editor doesn't bork client state We had a report of a situation like this borking a user at Google, but I can't reproduce it in a test. Let's at least backstop the issue with a test, so we don't accidentally introduce such a bug... Differential Revision: https://phab.mercurial-scm.org/D8391

rust-chg: silence warning about dated coding style Differential Revision: https://phab.mercurial-scm.org/D8384

rust-chg: update name of the server process This is a copy of updateprocname() of hgclient.c. At this point, rust-chg is basically functional. I did dogfooding for a couple of weeks in 2018. There are a few remaining tasks: a. loop detection by CHGINTERNALMARK b. forward unsupported commands (notably serve -d) to real hg c. better handling of early server exception d. modernize codebase (2018 edition, impl trait, async/await) For (d), we'll probably want to switch to async-std, but I'm thinking of upgrading to Tokio 0.2 as an intermediate step since process API isn't ported to async-std yet. I'm pretty sure future migration to async-std will be painless compared to the mass rewrite from futures-0.1 to 0.3. https://github.com/async-rs/async-std/issues/22 Differential Revision: https://phab.mercurial-scm.org/D8383

rust-chg: send client side umask to server This is equivalent to forwardumask() of hgclient.c. Differential Revision: https://phab.mercurial-scm.org/D8382

rust-chg: add config validation and process returned instructions This is the reimplementation of runinstructions() and main() in chg.c. In Rust version, we only pass in early arguments to the server as the locator doesn't know the full arguments. This should be fine since these arguments are just passed in to _earlyparseopts() and _parseconfig(), which means the server doesn't need full arguments. Another difference is the handling of the "exit <code>" instruction. In Rust version, we can simply reuse the connection instead of "exit(code)" as the command error isn't displayed yet. That's because the client-side stdio is not attached until the connection is validated. This behavior is cleaner than C, but it also means that the early server exception wouldn't be propagated to client because stderr isn't attached. So we might have to reconsider when to attach/detach the server stdio. Differential Revision: https://phab.mercurial-scm.org/D8381

rust-chg: collect server flags from command arguments This is the reimplementation of testsensitiveflag() and setcmdserverargs() of chg.c. Differential Revision: https://phab.mercurial-scm.org/D8380

rust-chg: add interface to run "validate" request Differential Revision: https://phab.mercurial-scm.org/D8379

rust-chg: add helper to parse instructions sent from server This is well structured version of runinstructions() of chg.c. Differential Revision: https://phab.mercurial-scm.org/D8378

tests: move verification closer to setup in test-copies-chain-merge.t I found this test case really hard to read because it requires scrolling back and forth between the setup and the verification. This patch moves them closer together. There was also duplicated descriptions for each test case (duplicated between setup and verification). They had already become out of sync. I picked the description from the verification, partly so the differences would be highlighted in this patch. Differential Revision: https://phab.mercurial-scm.org/D8377

tests: collect all branch creation in one place in test-copies-chain-merge.t I found this test case really hard to read because it requires scrolling back and forth between the setup and the verification. The next patch will move the verification close to the merge commits they test. This patch prepares for that by moving all branch creation first so they are separate from the merge commits (because many merge commits reuse the same branches). Differential Revision: https://phab.mercurial-scm.org/D8376

hgk: remove a "b" used on a kwargs expansion, the keys are strs Differential Revision: https://phab.mercurial-scm.org/D8385

rust-chg: send client-side environment variables to server This is also needed to run config validation. The "validate" request and its response handling will be implemented in the next batch. Differential Revision: https://phab.mercurial-scm.org/D8364

rust-chg: add helper to pack environment variables On my machine, "printenv | wc -c" says 3422. That's the only reason why the initial buffer capacity is set to 4kB. Differential Revision: https://phab.mercurial-scm.org/D8363

rust-chg: move set_current_dir() to Locator This is necessary to run config validation in proper environment. Differential Revision: https://phab.mercurial-scm.org/D8362

rust-chg: abort if server doesn't have required capabilities It's checked at Locator::connect() since we'll have to do more stuff in this function. Differential Revision: https://phab.mercurial-scm.org/D8361

rust-chg: spawn server process if not running This is the minimal reimplementation of gethgcmd(), execcmdserver(), retryconnectcmdserver(), and connectcmdserver() in chg.c. No config validation is implemented yet. And some Py3 workarounds would be missing as this is the code I wrote in 2018. Differential Revision: https://phab.mercurial-scm.org/D8360

rust-chg: update dependencies Differential Revision: https://phab.mercurial-scm.org/D8359

rust-chg: depend on released version of tokio-process Still it is futures-0.1 based. Differential Revision: https://phab.mercurial-scm.org/D8358

rust-chg: add brief comment about initial capacity of temp_sock_path() I don't know if it can be expressed as a compile-time constant, so it's a comment for now. About this series: This is quite old patches for rust-chg. I heard from Octobus people that there's a plan to do an experiment on merging hgcli + chg + some Rust?, so I decided to respin the rust-chg series. Maybe we'll rewrite the core to leverage the recent async/await functionality, but I want to first make my old patches in so the rust-chg can be a drop-in replacement for the chg of C. Compiler warnings will be removed later, and the codebase will be upgraded to the 2018 edition later. Differential Revision: https://phab.mercurial-scm.org/D8357

discovery: avoid wrongly saying there are nothing to pull We can get in a situation where a revision passed through `hg pull --rev REV` are available on the server, but not a descendant of the advertised server heads. For example the server could lying be during heads advertisement, to hide some pull request. Or obsolete/hidden content could be explicitly pulled. So in this case the lookup associated to `REV` returned successfully, but the normal discovery will find all advertised heads already known locally. This flip a special boolean `anyinc` that will prevent any fetch attempt, preventing `REV` to be pulled over. We add three line of code to detect this case and make sure a pull actually happens. My main target is to make some third party extensions happy (I expect the associated test to move upstream with the extension). However this fix already make some of the `infinitepush` test happier.

rebase: don't create merge when continuing rebase interrupted by old hg This fixes the bug described and demonstrated in the previous commit. It does so by practically undoing 8082a77cc3a2 (rebase: remove some redundant setting of dirstate parents, 2020-01-10). Differential Revision: https://phab.mercurial-scm.org/D8356

tests: demonstrate how continuing rebase after upgrade can result in merge If the user starts a rebase with an hg version before 9c9cfecd4600 (rebase: don't use rebased node as dirstate p2 (BC), 2020-01-10) and then runs into conflicts, they will be dropped out to the shell with the rebased node set as the dirstate's second parent. If they then upgrade to a later hg version, it will respect the dirstate's parents and will create a merge commit even if the user was rebasing a non-merge commit. Differential Revision: https://phab.mercurial-scm.org/D8355

manifest: also declare treemanifest as implementing imanifestdict It always has been intended to be, but it had a mismatched flags() method until the previous change. Differential Revision: https://phab.mercurial-scm.org/D8366

manifest: introduce new exception to signal unavailability of fastdelta() I've spent some time reflecting on this, and I think this is the best we can do in this API contract for now. This opens the door to adding treemanifest's implementation to the list of implementers of imanifestdict. Differential Revision: https://phab.mercurial-scm.org/D8365

copies-tests: remove spurious `]` in the template Fixing this typo happily impact all the tests output. Differential Revision: https://phab.mercurial-scm.org/D8258

merge with stable

debugcommands: fix typo in debuguigetpass

dagop: fix subsetparentswalker to set p1/p2 chains at merge revision The previous implementation was wrong because the '1'/'2' key would be appended at a fork revision. Since we traverse the graph from heads, a merge revision is actually a branching point, where the sort key must be generated.