{"thread":{"id":"47507","subject":"[ANNOUNCE] Git v2.16.0-rc0","startedAt":"2017-12-29T04:30:55Z","lastAt":"2018-06-15T07:26:25Z","messageCount":20,"participants":["Junio C Hamano","Kaartic Sivaraam","Paul Smith","Bryan Turner","Jonathan Nieder","Thomas Gummerer","Johannes Schindelin","Jeff King"],"isPatch":false,"patchVersion":null,"patchTotal":null},"messages":[{"id":"335515","messageId":"xmqqfu7ui2af.fsf@gitster.mtv.corp.google.com","threadId":"47507","inReplyTo":null,"subject":"[ANNOUNCE] Git v2.16.0-rc0","fromName":"Junio C Hamano","fromEmail":"gitster@pobox.com","sentAt":"2017-12-29T04:30:32Z","receivedAt":"2017-12-29T04:30:55Z","isPatch":false,"sender":{"key":"gitster@pobox.com","avatar":"https://avatars.githubusercontent.com/u/54884?v=4"},"body":"An early preview release Git v2.16.0-rc0 is now available for\ntesting at the usual places. It is comprised of 435 non-merge\ncommits since v2.15.0, contributed by 76 people, 22 of which are\nnew faces.\n\nThe tarballs are found at:\n\n https://www.kernel.org/pub/software/scm/git/testing/\n\nThe following public repositories all have a copy of the\n'v2.16.0-rc0' tag and the 'master' branch that the tag points at:\n\n url = https://kernel.googlesource.com/pub/scm/git/git\n url = git://repo.or.cz/alt-git.git\n url = https://github.com/gitster/git\n\nNew contributors whose contributions weren't in v2.15.0 are as follows.\nWelcome to the Git development community!\n\n Albert Astals Cid, Antoine Beaupré, Damien Marié, Daniel\n Bensoussan, Florian Klink, Gennady Kupava, Guillaume Castagnino,\n Haaris Mehmood, Hans Jerry Illikainen, Ingo Ruhnke, Jakub\n Bereżański, Jean Carlo Machado, J Wyman, Kevin, Łukasz\n Stelmach, Marius Paliga, Olga Telezhnaya, Rafael Ascensão,\n Robert Abel, Robert P. J. Day, Shuyu Wei, and Wei Shuyu.\n\nReturning contributors who helped this release are as follows.\nThanks for your continued support.\n\n Adam Dinwoodie, Ævar Arnfjörð Bjarmason, Alex Vandiver,\n Anders Kaseorg, Andrey Okoshkin, Ann T Ropea, Beat Bolli, Ben\n Peart, Brandon Williams, brian m. carlson, Carlos Martín Nieto,\n Charles Bailey, Christian Couder, Dennis Kaarsemaker, Derrick\n Stolee, Elijah Newren, Emily Xie, Eric Sunshine, Eric Wong,\n Heiko Voigt, Jacob Keller, Jameson Miller, Jean-Noel Avila,\n Jeff Hostetler, Jeff King, Johannes Schindelin, Jonathan Nieder,\n Jonathan Tan, Junio C Hamano, Kaartic Sivaraam, Kevin Daudt,\n Lars Schneider, Liam Beguin, Martin Ågren, Michael Haggerty,\n Nicolas Morey-Chaisemartin, Phil Hord, Phillip Wood, Pranit\n Bauva, Prathamesh Chavan, Ramsay Jones, Randall S. Becker,\n Rasmus Villemoes, René Scharfe, Simon Ruderich, Stefan Beller,\n Steffen Prohaska, Stephan Beyer, SZEDER Gábor, Thomas Braun,\n Thomas Gummerer, Todd Zullinger, Torsten Bögershausen, and\n W. Trevor King.\n\n----------------------------------------------------------------\n\nGit 2.16 Release Notes (draft)\n==============================\n\nBackward compatibility notes and other notable changes.\n\n * Use of an empty string as a pathspec element that is used for\n 'everything matches' is now an error.\n\n\nUpdates since v2.15\n-------------------\n\nUI, Workflows & Features\n\n * An empty string as a pathspec element that means \"everything\"\n i.e. 'git add \"\"', is now illegal. We started this by first\n deprecating and warning a pathspec that has such an element in\n 2.11 (Nov 2016).\n\n * A hook script that is set unexecutable is simply ignored. Git\n notifies when such a file is ignored, unless the message is\n squelched via advice.ignoredHook configuration.\n\n * \"git pull\" has been taught to accept \"--[no-]signoff\" option and\n pass it down to \"git merge\".\n\n * The \"--push-option=\" option to \"git push\" now defaults to a\n list of strings configured via push.pushOption variable.\n\n * \"gitweb\" checks if a directory is searchable with Perl's \"-x\"\n operator, which can be enhanced by using \"filetest 'access'\"\n pragma, which now we do.\n\n * \"git stash save\" has been deprecated in favour of \"git stash push\".\n\n * The set of paths output from \"git status --ignored\" was tied\n closely with its \"--untracked=\" option, but now it can be\n controlled more flexibly. Most notably, a directory that is\n ignored because it is listed to be ignored in the ignore/exclude\n mechanism can be handled differently from a directory that ends up\n to be ignored only because all files in it are ignored.\n\n * The remote-helper for talking to MediaWiki has been updated to\n truncate an overlong pagename so that \".mw\" suffix can still be\n added.\n\n * The remote-helper for talking to MediaWiki has been updated to\n work with mediawiki namespaces.\n\n * The \"--format=...\" option \"git for-each-ref\" takes learned to show\n the name of the 'remote' repository and the ref at the remote side\n that is affected for 'upstream' and 'push' via \"%(push:remotename)\"\n and friends.\n\n * Doc and message updates to teach users \"bisect view\" is a synonym\n for \"bisect visualize\".\n\n * \"git bisect run\" that did not specify any command to run used to go\n ahead and treated all commits to be tested as 'good'. This has\n been corrected by making the command error out.\n\n * The SubmittingPatches document has been converted to produce an\n HTML version via AsciiDoc/Asciidoctor.\n\n * We learned to talk to watchman to speed up \"git status\" and other\n operations that need to see which paths have been modified.\n\n * The \"diff\" family of commands learned to ignore differences in\n carriage return at the end of line.\n\n * Places that know about \"sendemail.to\", like documentation and shell\n completion (in contrib/) have been taught about \"sendemail.tocmd\",\n too.\n\n * \"git add --renormalize .\" is a new and safer way to record the fact\n that you are correcting the end-of-line convention and other\n \"convert_to_git()\" glitches in the in-repository data.\n\n * \"git branch\" and \"git checkout -b\" are now forbidden from creating\n a branch whose name is \"HEAD\".\n\n * \"git branch --list\" learned to show its output through the pager by\n default when the output is going to a terminal, which is controlled\n by the pager.branch configuration variable. This is similar to a\n recent change to \"git tag --list\".\n\n * \"git grep -W\", \"git diff -W\" and their friends learned a heuristic\n to extend a pre-context beyond the line that matches the \"function\n pattern\" (aka \"diff.*.xfuncname\") to include a comment block, if\n exists, that immediately precedes it.\n\n * \"git config --expiry-date gc.reflogexpire\" can read \"2.weeks\" from\n the configuration and report it as a timestamp, just like \"--int\"\n would read \"1k\" and report 1024, to help consumption by scripts.\n\n * The shell completion (in contrib/) learned that \"git pull\" can take\n the \"--autostash\" option.\n\n * The tagnames \"git log --decorate\" uses to annotate the commits can\n now be limited to subset of available refs with the two additional\n options, --decorate-refs[-exclude]=.\n\n * \"git grep\" compiled with libpcre2 sometimes triggered a segfault,\n which is being fixed.\n\n * \"git send-email\" tries to see if the sendmail program is available\n in /usr/lib and /usr/sbin; extend the list of locations to be\n checked to also include directories on $PATH.\n\n * \"git diff\" learned, \"--anchored\", a variant of the \"--patience\"\n algorithm, to which the user can specify which 'unique' line to be\n used as anchoring points.\n\n * The way \"git worktree add\" determines what branch to create from\n where and checkout in the new worktree has been updated a bit.\n\n * Ancient part of codebase still shows dots after an abbreviated\n object name just to show that it is not a full object name, but\n these ellipses are confusing to people who newly discovered Git\n who are used to seeing abbreviated object names and find them\n confusing with the range syntax.\n\n * With a configuration variable rebase.abbreviateCommands set,\n \"git rebase -i\" produces the todo list with a single-letter\n command names.\n\n * \"git worktree add\" learned to run the post-checkout hook, just like\n \"git checkout\" does, after the initial checkout.\n\n * \"git svn\" has been updated to strip CRs in the commit messages, as\n recent versions of Subversion rejects them.\n\n * \"git imap-send\" did not correctly quote the folder name when\n making a request to the server, which has been corrected.\n\n * Error messages from \"git rebase\" have been somewhat cleaned up.\n\n * Git has been taught to support an https:// URL used for http.proxy\n when using recent versions of libcurl.\n\n * \"git merge\" learned to pay attention to merge.verifySignatures\n configuration variable and pretend as if '--verify-signatures'\n option was given from the command line.\n\n * \"git describe\" was taught to dig trees deeper to find a\n : that refers to a given blob object.\n\n\nPerformance, Internal Implementation, Development Support etc.\n\n * An earlier update made it possible to use an on-stack in-core\n lockfile structure (as opposed to having to deliberately leak an\n on-heap one). Many codepaths have been updated to take advantage\n of this new facility.\n\n * Calling cmd_foo() as if it is a general purpose helper function is\n a no-no. Correct two instances of such to set an example.\n\n * We try to see if somebody runs our test suite with a shell that\n does not support \"local\" like bash/dash does.\n\n * An early part of piece-by-piece rewrite of \"git bisect\" in C.\n\n * GSoC to piece-by-piece rewrite \"git submodule\" in C.\n\n * Optimize the code to find shortest unique prefix of object names.\n\n * Pathspec-limited revision traversal was taught not to keep finding\n unneeded differences once it knows two trees are different inside\n given pathspec.\n\n * Conversion from uchar[20] to struct object_id continues.\n\n * Code cleanup.\n\n * A single-word \"unsigned flags\" in the diff options is being split\n into a structure with many bitfields.\n\n * TravisCI build updates.\n\n * Parts of a test to drive the long-running content filter interface\n has been split into its own module, hopefully to eventually become\n reusable.\n\n * Drop (perhaps overly cautious) sanity check before using the index\n read from the filesystem at runtime.\n\n * The build procedure has been taught to avoid some unnecessary\n instability in the build products.\n\n * A new mechanism to upgrade the wire protocol in place is proposed\n and demonstrated that it works with the older versions of Git\n without harming them.\n\n * An infrastructure to define what hash function is used in Git is\n introduced, and an effort to plumb that throughout various\n codepaths has been started.\n\n * The code to iterate over loose object files got optimized.\n\n * An internal function that was left for backward compatibility has\n been removed, as there is no remaining callers.\n\n * Historically, the diff machinery for rename detection had a\n hardcoded limit of 32k paths; this is being lifted to allow users\n trade cycles with a (possibly) easier to read result.\n\n * The tracing infrastructure has been optimized for cases where no\n tracing is requested.\n\n * In preparation for implementing narrow/partial clone, the object\n walking machinery has been taught a way to tell it to \"filter\" some\n objects from enumeration.\n\n * A few structures and variables that are implementation details of\n the decorate API have been renamed and then the API got documented\n better.\n\n * Assorted updates for TravisCI integration.\n (merge 4f26366679 sg/travis-fixes later to maint).\n\n * Introduce a helper to simplify code to parse a common pattern that\n expects either \"--key\" or \"--key=\".\n\n * \"git version --build-options\" learned to report the host CPU and\n the exact commit object name the binary was built from.\n\nAlso contains various documentation updates and code clean-ups.\n\n\nFixes since v2.15\n-----------------\n\n * \"auto\" as a value for the columnar output configuration ought to\n judge \"is the output consumed by humans?\" with the same criteria as\n \"auto\" for coloured output configuration, i.e. either the standard\n output stream is going to tty, or a pager is in use. We forgot the\n latter, which has been fixed.\n\n * The experimental \"color moved lines differently in diff output\"\n feature was buggy around \"ignore whitespace changes\" edges, which\n has been corrected.\n\n * Instead of using custom line comparison and hashing functions to\n implement \"moved lines\" coloring in the diff output, use the pair\n of these functions from lower-layer xdiff/ code.\n\n * Some codepaths did not check for errors when asking what branch the\n HEAD points at, which have been fixed.\n\n * \"git commit\", after making a commit, did not check for errors when\n asking on what branch it made the commit, which has been corrected.\n\n * \"git status --ignored -u\" did not stop at a working tree of a\n separate project that is embedded in an ignored directory and\n listed files in that other project, instead of just showing the\n directory itself as ignored.\n\n * A broken access to object databases in recent update to \"git grep\n --recurse-submodules\" has been fixed.\n\n * A recent regression in \"git rebase -i\" that broke execution of git\n commands from subdirectories via \"exec\" instruction has been fixed.\n\n * A (possibly flakey) test fix.\n\n * \"git check-ref-format --branch @{-1}\" bit a \"BUG()\" when run\n outside a repository for obvious reasons; clarify the documentation\n and make sure we do not even try to expand the at-mark magic in\n such a case, but still call the validation logic for branch names.\n\n * \"git fetch --recurse-submodules\" now knows that submodules can be\n moved around in the superproject in addition to getting updated,\n and finds the ones that need to be fetched accordingly.\n\n * Command line completion (in contrib/) update.\n\n * Description of blame.{showroot,blankboundary,showemail,date}\n configuration variables have been added to \"git config --help\".\n\n * After an error from lstat(), diff_populate_filespec() function\n sometimes still went ahead and used invalid data in struct stat,\n which has been fixed.\n\n * UNC paths are also relevant in Cygwin builds and they are now\n tested just like Mingw builds.\n\n * Correct start-up sequence so that a repository could be placed\n immediately under the root directory again (which was broken at\n around Git 2.13).\n\n * The credential helper for libsecret (in contrib/) has been improved\n to allow possibly prompting the end user to unlock secrets that are\n currently locked (otherwise the secrets may not be loaded).\n\n * MinGW updates.\n\n * Error checking in \"git imap-send\" for empty response has been\n improved.\n\n * Recent update to the refs infrastructure implementation started\n rewriting packed-refs file more often than before; this has been\n optimized again for most trivial cases.\n\n * Some error messages did not quote filenames shown in it, which have\n been fixed.\n\n * \"git rebase -i\" recently started misbehaving when a submodule that\n is configured with 'submodule..ignore' is dirty; this has\n been corrected.\n\n * Building with NO_LIBPCRE1_JIT did not disable it, which has been fixed.\n\n * We used to add an empty alternate object database to the system\n that does not help anything; it has been corrected.\n\n * Doc update around use of \"format-patch --subject-prefix\" etc.\n\n * A fix for an ancient bug in \"git apply --ignore-space-change\" codepath.\n\n * Clarify and enhance documentation for \"merge-base --fork-point\", as\n it was clear what it computed but not why/what for.\n\n * A few scripts (both in production and tests) incorrectly redirected\n their error output. These have been corrected.\n\n * \"git notes\" sent its error message to its standard output stream,\n which was corrected.\n\n * The three-way merge performed by \"git cherry-pick\" was confused\n when a new submodule was added in the meantime, which has been\n fixed (or \"papered over\").\n\n * The sequencer machinery (used by \"git cherry-pick A..B\", and \"git\n rebase -i\", among other things) would have lost a commit if stopped\n due to an unlockable index file, which has been fixed.\n\n * \"git apply --inaccurate-eof\" when used with \"--ignore-space-change\"\n triggered an internal sanity check, which has been fixed.\n\n * Command line completion (in contrib/) has been taught about the\n \"--copy\" option of \"git branch\".\n\n * When \"git rebase\" prepared an mailbox of changes and fed it to \"git\n am\" to replay them, it was confused when a stray \"From \" happened\n to be in the log message of one of the replayed changes. This has\n been corrected.\n\n * There was a recent semantic mismerge in the codepath to write out a\n section of a configuration section, which has been corrected.\n\n * Mentions of \"git-rebase\" and \"git-am\" (dashed form) still remained\n in end-user visible strings emitted by the \"git rebase\" command;\n they have been corrected.\n\n * Contrary to the documentation, \"git pull -4/-6 other-args\" did not\n ask the underlying \"git fetch\" to go over IPv4/IPv6, which has been\n corrected.\n\n * \"git checkout --recursive\" may overwrite and rewind the history of\n the branch that happens to be checked out in submodule\n repositories, which might not be desirable. Detach the HEAD but\n still allow the recursive checkout to succeed in such a case.\n (merge 57f22bf997 sb/submodule-recursive-checkout-detach-head later to maint).\n\n * \"git branch --set-upstream\" has been deprecated and (sort of)\n removed, as \"--set-upstream-to\" is the preferred one these days.\n The documentation still had \"--set-upstream\" listed on its\n synopsis section, which has been corrected.\n (merge a060f3d3d8 tz/branch-doc-remove-set-upstream later to maint).\n\n * Internally we use 0{40} as a placeholder object name to signal the\n codepath that there is no such object (e.g. the fast-forward check\n while \"git fetch\" stores a new remote-tracking ref says \"we know\n there is no 'old' thing pointed at by the ref, as we are creating\n it anew\" by passing 0{40} for the 'old' side), and expect that a\n codepath to locate an in-core object to return NULL as a sign that\n the object does not exist. A look-up for an object that does not\n exist however is quite costly with a repository with large number\n of packfiles. This access pattern has been optimized.\n (merge 87b5e236a1 jk/fewer-pack-rescan later to maint).\n\n * In addition to \"git stash -m message\", the command learned to\n accept \"git stash -mmessage\" form.\n (merge 5675473fcb ph/stash-save-m-option-fix later to maint).\n\n * @{-N} in \"git checkout @{-N}\" may refer to a detached HEAD state,\n but the documentation was not clear about it, which has been fixed.\n (merge 75ce149575 ks/doc-checkout-previous later to maint).\n\n * A regression in the progress eye-candy was fixed.\n (merge 9c5951cacf jk/progress-delay-fix later to maint).\n\n * The code internal to the recursive merge strategy was not fully\n prepared to see a path that is renamed to try overwriting another\n path that is only different in case on case insensitive systems.\n This does not matter in the current code, but will start to matter\n once the rename detection logic starts taking hints from nearby\n paths moving to some directory and moves a new path along with them.\n (merge 4cba2b0108 en/merge-recursive-icase-removal later to maint).\n\n * An v2.12-era regression in pathspec match logic, which made it look\n into submodule tree even when it is not desired, has been fixed.\n (merge eef3df5a93 bw/pathspec-match-submodule-boundary later to maint).\n\n * Amending commits in git-gui broke the author name that is non-ascii\n due to incorrect enconding conversion.\n\n * Recent update to the submodule configuration code broke \"diff-tree\"\n by accidentally stopping to read from the index upfront.\n (merge fd66bcc31f bw/submodule-config-cleanup later to maint).\n\n * Git shows a message to tell the user that it is waiting for the\n user to finish editing when spawning an editor, in case the editor\n opens to a hidden window or somewhere obscure and the user gets\n lost.\n (merge abfb04d0c7 ls/editor-waiting-message later to maint).\n\n * The \"safe crlf\" check incorrectly triggered for contents that does\n not use CRLF as line endings, which has been corrected.\n (merge 649f1f0948 tb/check-crlf-for-safe-crlf later to maint).\n\n * \"git clone --shared\" to borrow from a (secondary) worktree did not\n work, even though \"git clone --local\" did. Both are now accepted.\n (merge b3b05971c1 es/clone-shared-worktree later to maint).\n\n * The build procedure now allows not just the repositories but also\n the refs to be used to take pre-formatted manpages and html\n documents to install.\n (merge 65289e9dcd rb/quick-install-doc later to maint).\n\n * Update the shell prompt script (in contrib/) to strip trailing CR\n from strings read from various \"state\" files.\n (merge 041fe8fc83 ra/prompt-eread-fix later to maint).\n\n * Other minor doc, test and build updates and code cleanups.\n (merge 1a1fc2d5b5 rd/man-prune-progress later to maint).\n (merge 0ba014035a rd/man-reflog-add-n later to maint).\n (merge e54b63359f rd/doc-notes-prune-fix later to maint).\n (merge ff4c9b413a sp/doc-info-attributes later to maint).\n (merge 7db2cbf4f1 jc/receive-pack-hook-doc later to maint).\n (merge 5a0526264b tg/t-readme-updates later to maint).\n (merge 5e83cca0b8 jk/no-optional-locks later to maint).\n (merge 826c778f7c js/hashmap-update-sample later to maint).\n (merge 176b2d328c sg/setup-doc-update later to maint).\n (merge 1b09073514 rs/am-builtin-leakfix later to maint).\n (merge addcf6cfde rs/fmt-merge-msg-string-leak-fix later to maint).\n (merge c3ff8f6c14 rs/strbuf-read-once-reset-length later to maint).\n (merge 6b0eb884f9 db/doc-workflows-neuter-the-maintainer later to maint).\n (merge 8c87bdfb21 jk/cvsimport-quoting later to maint).\n (merge 176cb979fe rs/fmt-merge-msg-leakfix later to maint).\n (merge 5a03360e73 tb/delimit-pretty-trailers-args-with-comma later to maint).\n (merge d0e6326026 ot/pretty later to maint).\n (merge 44103f4197 sb/test-helper-excludes later to maint).\n (merge 170078693f jt/transport-no-more-rsync later to maint).\n (merge c07b3adff1 bw/path-doc later to maint).\n (merge bf9d7df950 tz/lib-git-svn-svnserve-tests later to maint).\n (merge dec366c9a8 sr/http-sslverify-config-doc later to maint).\n\n----------------------------------------------------------------\n\nChanges since v2.15.0 are as follows:\n\nAdam Dinwoodie (3):\n t5580: add Cygwin support\n rebase -i: fix comment typo\n doc/SubmittingPatches: correct subject guidance\n\nAlbert Astals Cid (1):\n completion: add --autostash and --no-autostash to pull\n\nAlex Vandiver (7):\n fsmonitor: read entirety of watchman output\n fsmonitor: set the PWD to the top of the working tree\n fsmonitor: don't bother pretty-printing JSON from watchman\n fsmonitor: document GIT_TRACE_FSMONITOR\n fsmonitor: delay updating state until after split index is merged\n fsmonitor: read from getcwd(), not the PWD environment variable\n fsmonitor: store fsmonitor bitmap before splitting index\n\nAnders Kaseorg (1):\n git-gui: sort entries in optimized tclIndex\n\nAndrey Okoshkin (3):\n commit: check result of resolve_ref_unsafe\n diff: fix lstat() error handling in diff_populate_filespec()\n merge-recursive: check GIT_MERGE_VERBOSITY only once\n\nAnn T Ropea (7):\n Documentation: revisions: fix typo: \"three dot\" ---> \"three-dot\" (in line with \"two-dot\").\n Documentation: user-manual: limit usage of ellipsis\n print_sha1_ellipsis: introduce helper\n checkout: describe_detached_head: remove ellipsis after committish\n t4013: prepare for upcoming \"diff --raw --abbrev\" output format change\n diff: diff_aligned_abbrev: remove ellipsis after abbreviated SHA-1 value\n t4013: test new output from diff --abbrev --raw\n\nAntoine Beaupré (6):\n remote-mediawiki: limit filenames to legal\n remote-mediawiki: show known namespace choices on failure\n remote-mediawiki: skip virtual namespaces\n remote-mediawiki: support fetching from (Main) namespace\n remote-mediawiki: process namespaces in order\n remote-mediawiki: show progress while fetching namespaces\n\nBeat Bolli (1):\n contrib/git-jump: allow to configure the grep command\n\nBen Peart (16):\n bswap: add 64 bit endianness helper get_be64\n preload-index: add override to enable testing preload-index\n update-index: add a new --force-write-index option\n fsmonitor: teach git to optionally utilize a file system monitor to speed up detecting new or changed files.\n fsmonitor: add documentation for the fsmonitor extension.\n ls-files: Add support in ls-files to display the fsmonitor valid bit\n update-index: add fsmonitor support to update-index\n fsmonitor: add a test tool to dump the index extension\n split-index: disable the fsmonitor extension when running the split index test\n fsmonitor: add test cases for fsmonitor extension\n fsmonitor: add a sample integration script for Watchman\n fsmonitor: add a performance test\n fsmonitor: MINGW support for watchman integration\n fsmonitor: preserve utf8 filenames in fsmonitor-watchman log\n read_index_from(): speed index loading by skipping verification of the entry order\n fsmonitor: simplify determining the git worktree under Windows\n\nBrandon Williams (24):\n pkt-line: add packet_write function\n protocol: introduce protocol extension mechanisms\n daemon: recognize hidden request arguments\n upload-pack, receive-pack: introduce protocol version 1\n connect: teach client to recognize v1 server response\n connect: tell server that the client understands v1\n http: tell server that the client understands v1\n i5700: add interop test for protocol transition\n ssh: introduce a 'simple' ssh variant\n add, reset: use DIFF_OPT_SET macro to set a diff flag\n diff: convert flags to be stored in bitfields\n diff: add flag to indicate textconv was set via cmdline\n diff: remove touched flags\n diff: remove DIFF_OPT_TST macro\n diff: remove DIFF_OPT_SET macro\n diff: remove DIFF_OPT_CLR macro\n diff: make struct diff_flags members lowercase\n wt-status: actually ignore submodules when requested\n pathspec: only match across submodule boundaries when requested\n diff-tree: read the index so attribute checks work in bare repositories\n submodule: convert stage_updated_gitmodules to take a struct index_state\n submodule: used correct index in is_staging_gitmodules_ok\n submodule: convert get_next_submodule to not rely on the_index\n path: document path functions\n\nCarlos Martín Nieto (1):\n diff: --indent-heuristic is no longer experimental\n\nCharles Bailey (2):\n t4201: make use of abbreviation in the test more robust\n grep: fix NO_LIBPCRE1_JIT to fully disable JIT\n\nChristian Couder (24):\n perf/run: add '--config' option to the 'run' script\n perf/run: add get_var_from_env_or_config()\n perf/run: add GIT_PERF_DIRS_OR_REVS\n perf/run: add calls to get_var_from_env_or_config()\n perf/run: add get_subsections()\n perf/run: update get_var_from_env_or_config() for subsections\n perf/run: add run_subsection()\n perf/run: show name of rev being built\n perf: store subsection results in \"test-results/$GIT_PERF_SUBSECTION/\"\n t0021/rot13-filter: fix list comparison\n t0021/rot13-filter: refactor packet reading functions\n t0021/rot13-filter: improve 'if .. elsif .. else' style\n t0021/rot13-filter: improve error message\n t0021/rot13-filter: add packet_initialize()\n t0021/rot13-filter: refactor checking final lf\n t0021/rot13-filter: add capability functions\n Git/Packet.pm: extract parts of t0021/rot13-filter.pl for reuse\n Git/Packet: clarify that packet_required_key_val_read allows EOF\n Git/Packet.pm: use 'if' instead of 'unless'\n list-objects-filter-options: fix 'keword' typo in comment\n git-compat-util: introduce skip_to_optional_arg()\n index-pack: use skip_to_optional_arg()\n diff: use skip_to_optional_arg()\n diff: use skip_to_optional_arg_default()\n\nDamien Marié (1):\n run-command: add hint when a hook is ignored\n\nDaniel Bensoussan (1):\n doc: reword gitworkflows.txt for neutrality\n\nDennis Kaarsemaker (1):\n credential-libsecret: unlock locked secrets\n\nDerrick Stolee (5):\n p4211-line-log.sh: add log --online --raw --parents perf test\n sha1_name: unroll len loop in find_unique_abbrev_r()\n sha1_name: parse less while finding common prefix\n sha1_name: minimize OID comparisons during disambiguation\n sha1_file: use strbuf_add() instead of strbuf_addf()\n\nElijah Newren (7):\n sequencer: warn when internal merge may be suboptimal due to renameLimit\n merge-recursive: handle addition of submodule on our side of history\n progress: fix progress meters when dealing with lots of work\n diff: remove silent clamp of renameLimit\n sequencer: show rename progress during cherry picks\n merge-recursive: ignore_case shouldn't reject intentional removals\n strbuf: remove unused stripspace function alias\n\nEmily Xie (1):\n pathspec: die on empty strings as pathspec\n\nEric Sunshine (3):\n worktree: invoke post-checkout hook (unless --no-checkout)\n clone: support 'clone --shared' from a worktree\n version --build-options: also report host CPU\n\nEric Wong (2):\n rebase: use mboxrd format to avoid split errors\n git-svn: convert CRLF to LF in commit message to SVN\n\nFlorian Klink (1):\n git-send-email: honor $PATH for sendmail binary\n\nGennady Kupava (2):\n trace: remove trace key normalization\n trace: improve performance while category is disabled\n\nGuillaume Castagnino (1):\n gitweb: use filetest to allow ACLs\n\nHaaris Mehmood (1):\n config: add --expiry-date\n\nHans Jerry Illikainen (2):\n merge: add config option for verifySignatures\n t: add tests for pull --verify-signatures\n\nHeiko Voigt (3):\n fetch: add test to make sure we stay backwards compatible\n implement fetching of moved submodules\n submodule: simplify decision tree whether to or not to fetch\n\nIngo Ruhnke (1):\n remote-mediawiki: allow fetching namespaces with spaces\n\nJ Wyman (1):\n for-each-ref: let upstream/push report the remote ref name\n\nJacob Keller (2):\n sequencer: pass absolute GIT_DIR to exec commands\n diff: add tests for --relative without optional prefix value\n\nJakub Bereżański (2):\n t0302: check helper can handle empty credentials\n wincred: handle empty username/password correctly\n\nJameson Miller (4):\n status: add option to show ignored files differently\n status: report matching ignored and normal untracked\n status: document options to show matching ignored files\n status: test ignored modes\n\nJean Carlo Machado (1):\n fix typos in 2.15.0 release notes\n\nJean-Noel Avila (1):\n submodule--helper.c: i18n: add a missing space in message\n\nJeff Hostetler (9):\n dir: allow exclusions from blob in addition to file\n oidmap: add oidmap iterator methods\n oidset: add iterator methods to oidset\n list-objects: filter objects in traverse_commit_list\n rev-list: add list-objects filtering support\n pack-objects: add list-objects filtering\n list-objects-filter-options: support --no-filter\n rev-list: support --no-filter argument\n partial-clone: design doc\n\nJeff King (22):\n revision: quit pruning diff more quickly when possible\n t4015: refactor --color-moved whitespace test\n t4015: check \"negative\" case for \"-w --color-moved\"\n t4015: test the output of \"diff --color-moved -b\"\n diff: fix whitespace-skipping with --color-moved\n diff: handle NULs in get_string_hash()\n test-ref-store: avoid passing NULL to printf\n remote: handle broken symrefs\n log: handle broken HEAD in decoration check\n worktree: handle broken symrefs in find_shared_symref()\n setup: avoid double slashes when looking for HEAD\n link_alt_odb_entries: make empty input a noop\n git-jump: give contact instructions in the README\n p5550: factor out nonsense-pack creation\n t/perf/lib-pack: use fast-import checkpoint to create packs\n p5551: add a script to test fetch pack-dir rescans\n everything_local: use \"quick\" object existence check\n sha1_file: fast-path null sha1 as a missing object\n git-status.txt: mention --no-optional-locks\n progress: set default delay threshold to 100%, not 0%\n docs/pretty-formats: mention commas in %(trailers) syntax\n cvsimport: apply shell-quoting regex globally\n\nJohannes Schindelin (9):\n for-each-ref: let upstream/push optionally report the remote name\n status: do not get confused by submodules in excluded directories\n mingw: include the full version information in the resources\n mingw: add experimental feature to redirect standard handles\n mingw: optionally redirect stderr/stdout via the same handle\n mingw: document the standard handle redirection\n for-each-ref: test :remotename and :remoteref\n hashmap: adjust documentation to reflect reality\n version --build-options: report commit, too, if possible\n\nJonathan Nieder (10):\n ssh test: make copy_ssh_wrapper_as clean up after itself\n connect: move no_fork fallback to git_tcp_connect\n connect: split git:// setup into a separate function\n connect: split ssh command line options into separate function\n connect: split ssh option computation to its own function\n ssh: 'auto' variant to select between 'ssh' and 'simple'\n ssh: 'simple' variant does not support -4/-6\n ssh: 'simple' variant does not support --port\n connect: correct style of C-style comment\n generate-cmdlist: avoid non-deterministic output\n\nJonathan Tan (10):\n connect: in ref advertisement, shallows are last\n Documentation: document Extra Parameters\n Tests: clean up and document submodule helpers\n Tests: clean up submodule recursive helpers\n diff: support anchoring line(s)\n diffcore-rename: make diff-tree -l0 mean -l\n decorate: clean up and document API\n transport: remove unused \"push\" in vtable\n clone, fetch: remove redundant transport check\n transport: make transport vtable more private\n\nJunio C Hamano (37):\n t0027: do not use an empty string as a pathspec element\n describe: do not use cmd_*() as a subroutine\n merge-ours: do not use cmd_*() as a subroutine\n branch: streamline \"attr_only\" handling in validate_new_branchname()\n branch: split validate_new_branchname() into two\n t5601: rm the target file of cp that could still be executing\n check-ref-format --branch: do not expand @{...} outside repository\n check-ref-format --branch: strip refs/heads/ using skip_prefix\n check-ref-format doc: --branch validates and expands \n column: do not include pager.c\n xdiff: reassign xpparm_t.flags bits\n The first batch for 2.16\n RelNotes: the second batch post 2.15 comes\n diff: --ignore-cr-at-eol\n merge-base --fork-point doc: clarify the example and failure modes\n RelNotes: the third batch for 2.16\n branch: correctly reject refs/heads/{-dash,HEAD}\n Start preparation for 2.15.1\n RelNotes: the fourth batch for 2.16\n Almost ready for 2.15.1\n RelNotes: the fifth batch for 2.16\n hooks doc: clarify when receive-pack invokes its hooks\n A bit more fixes for 2.15.1\n RelNotes: the sixth batch for 2.16\n Git 2.15.1\n RelNotes: the seventh batch\n t2020: test variations that matter\n Prepare for 2.15.2\n RelNotes: the eighth batch\n diff: use skip_to_optional_arg_default() in parsing --relative\n t4045: reindent to make helpers readable\n RelNotes: the ninth batch\n RelNotes: the tenth batch\n t5573, t7612: clean up after unexpected success of 'pull' and 'merge'\n sequencer.c: drop 'const' from function return type\n RelNotes: the eleventh batch\n Git 2.16-rc0\n\nKaartic Sivaraam (12):\n mailmap: use Kaartic Sivaraam's new address\n builtin/branch: remove redundant check for HEAD\n git-rebase: clean up dashed-usages in messages\n Doc/checkout: checking out using @{-N} can lead to detached state\n branch: improve documentation and naming of create_branch() parameters\n branch: group related arguments of create_branch()\n branch: update warning message shown when copying a misnamed branch\n builtin/branch: strip refs/heads/ using skip_prefix\n Doc/check-ref-format: clarify information about @{-N} syntax\n rebase: consistently use branch_name variable\n rebase: distinguish user input by quoting it\n rebase: rebasing can also be done when HEAD is detached\n\nKevin (1):\n remote-mediawiki: add namespace support\n\nKevin Daudt (1):\n column: show auto columns when pager is active\n\nLars Schneider (3):\n refactor \"dumb\" terminal determination\n progress: drop delay-threshold code\n launch_editor(): indicate that Git waits for user input\n\nLiam Beguin (9):\n Documentation: move rebase.* configs to new file\n Documentation: use preferred name for the 'todo list' script\n rebase -i: set commit to null in exec commands\n rebase -i: refactor transform_todo_ids\n rebase -i: replace reference to sha1 with oid\n rebase -i: update functions to use a flags parameter\n rebase -i -x: add exec commands via the rebase--helper\n rebase -i: learn to abbreviate command names\n t3404: add test case for abbreviated commands\n\nMarius Paliga (1):\n builtin/push.c: add push.pushOption config\n\nMartin Ågren (22):\n sha1_file: do not leak `lock_file`\n treewide: prefer lockfiles on the stack\n lockfile: fix documentation on `close_lock_file_gently()`\n tempfile: fix documentation on `delete_tempfile()`\n checkout-index: simplify locking logic\n cache-tree: simplify locking logic\n apply: move lockfile into `apply_state`\n apply: remove `newfd` from `struct apply_state`\n cache.h: document `write_locked_index()`\n read-cache: drop explicit `CLOSE_LOCK`-flag\n read-cache: leave lock in right state in `write_locked_index()`\n read_cache: roll back lock in `update_index_if_able()`\n grep: take the read-lock when adding a submodule\n bisect: change calling-convention of `find_bisection()`\n bisect: fix memory leak in `find_bisection()`\n bisect: fix off-by-one error in `best_bisection_sorted()`\n bisect: fix memory leak when returning best element\n builtin/merge-base: free commit lists\n reduce_heads: fix memory leaks\n t7006: add tests for how git branch paginates\n branch: respect `pager.branch` in list-mode only\n branch: change default of `pager.branch` to \"on\"\n\nMichael Haggerty (12):\n t1409: check that `packed-refs` is not rewritten unnecessarily\n files-backend: don't rewrite the `packed-refs` file unnecessarily\n t0000: check whether the shell supports the \"local\" keyword\n files_transaction_prepare(): don't leak flags to packed transaction\n prune_ref(): call `ref_transaction_add_update()` directly\n ref_transaction_update(): die on disallowed flags\n ref_transaction_add_update(): remove a check\n refs: tidy up and adjust visibility of the `ref_update` flags\n refs: rename constant `REF_NODEREF` to `REF_NO_DEREF`\n refs: rename constant `REF_ISPRUNING` to `REF_IS_PRUNING`\n write_packed_entry(): take `object_id` arguments\n refs: update some more docs to use \"oid\" rather than \"sha1\"\n\nNicolas Morey-Chaisemartin (1):\n imap-send: URI encode server folder\n\nOlga Telezhnaya (2):\n format: create pretty.h file\n format: create docs for pretty.h\n\nPhil Hord (2):\n doc: prefer 'stash push' over 'stash save'\n stash: learn to parse -m/--message like commit does\n\nPhillip Wood (2):\n config: avoid \"write_in_full(fd, buf, len) != len\" pattern\n sequencer: reschedule pick if index can't be locked\n\nPranit Bauva (6):\n bisect--helper: use OPT_CMDMODE instead of OPT_BOOL\n bisect--helper: rewrite `check_term_format` shell function in C\n bisect--helper: `write_terms` shell function in C\n bisect--helper: `bisect_clean_state` shell function in C\n t6030: explicitly test for bisection cleanup\n bisect--helper: `is_expected_rev` & `check_expected_revs` shell function in C\n\nPrathamesh Chavan (3):\n submodule--helper: introduce get_submodule_displaypath()\n submodule--helper: introduce for_each_listed_submodule()\n submodule: port submodule subcommand 'status' from shell to C\n\nRafael Ascensão (1):\n log: add option to choose which refs to decorate\n\nRamsay Jones (1):\n repository: fix a sparse 'using integer as NULL pointer' warning\n\nRandall S. Becker (1):\n install-doc-quick: allow specifying what ref to install\n\nRasmus Villemoes (2):\n Documentation/config: add sendemail.tocmd to list preceding \"See git-send-email(1)\"\n completion: add git config sendemail.tocmd\n\nRené Scharfe (21):\n notes: move hex_to_bytes() to hex.c and export it\n http-push: use hex_to_bytes()\n sha1_file: use hex_to_bytes()\n sequencer: factor out rewrite_file()\n sequencer: use O_TRUNC to truncate files\n imap-send: handle NULL return of next_arg()\n imap-send: handle missing response codes gracefully\n apply: avoid out-of-bounds access in fuzzy_matchlines()\n apply: update line lengths for --inaccurate-eof\n config: flip return value of write_section()\n t4051: add test for comments preceding function lines\n xdiff: factor out is_func_rec()\n xdiff: show non-empty lines before functions with -W\n t7810: improve check of -W with user-defined function lines\n grep: update boundary variable for pre-context\n grep: show non-empty lines before functions with -W\n am: release strbuf after use in split_mail_mbox()\n fmt-merge-msg: avoid leaking strbuf in shortlog()\n strbuf: release memory on read error in strbuf_read_once()\n transport-helper: plug strbuf and string_list leaks\n p7519: improve check for prerequisite WATCHMAN\n\nRobert Abel (2):\n git-prompt: make __git_eread intended use explicit\n git-prompt: fix reading files with windows line endings\n\nRobert P. J. Day (4):\n bisect: mention \"view\" as an alternative to \"visualize\"\n doc: add missing \"-n\" (dry-run) option to reflog man page\n prune: add \"--progress\" to man page and usage msg\n notes: correct 'git notes prune' options to '[-n] [-v]'\n\nSZEDER Gábor (7):\n travis-ci: fix running P4 and Git LFS tests in Linux build jobs\n travis-ci: don't build Git for the static analysis job\n setup.c: fix comment about order of .git directory discovery\n travis-ci: introduce a $jobname variable for 'ci/*' scripts\n travis-ci: move setting environment variables to 'ci/lib-travisci.sh'\n travis-ci: set GIT_TEST_HTTPD in 'ci/lib-travisci.sh'\n travis-ci: use 'set -x' in 'ci/*' scripts for extra tracing output\n\nShuyu Wei (1):\n pull: pass -4/-6 option to 'git fetch'\n\nSimon Ruderich (3):\n sequencer.c: check return value of close() in rewrite_file()\n wrapper.c: consistently quote filenames in error messages\n config: document default value of http.sslVerify\n\nStefan Beller (15):\n recursive submodules: detach HEAD from new state\n Documentation/checkout: clarify submodule HEADs to be detached\n xdiff-interface: export comparing and hashing strings\n diff.c: get rid of duplicate implementation\n t6120: fix typo in test name\n list-objects.c: factor out traverse_trees_and_blobs\n config: document blame configuration\n t/3512: demonstrate unrelated submodule/file conflict as cherry-pick failure\n revision.h: introduce blob/tree walking in order of the commits\n builtin/describe.c: rename `oid` to avoid variable shadowing\n builtin/describe.c: print debug statements earlier\n builtin/describe.c: factor out describe_commit\n Documentation/git-clone: improve description for submodule recursing\n t/helper: ignore everything but sources\n builtin/describe.c: describe a blob\n\nSteffen Prohaska (1):\n doc: Mention info/attributes in gitrepository-layout\n\nStephan Beyer (1):\n bisect run: die if no command is given\n\nThomas Braun (1):\n completion: add remaining flags to checkout\n\nThomas Gummerer (11):\n stash: replace \"git stash save\" with \"git stash push\" in the documentation\n stash: mark \"git stash save\" deprecated in the man page\n stash: remove now superfluos help for \"stash push\"\n checkout: factor out functions to new lib file\n worktree: add can be created from any commit-ish\n worktree: add --[no-]track option to the add subcommand\n worktree: make add dwim\n t/README: remove mention of adding copyright notices\n t/README: document test_cmp_rev\n worktree: add --guess-remote flag to add subcommand\n add worktree.guessRemote config option\n\nTodd Zullinger (11):\n Replace Free Software Foundation address in license notices\n Replace Free Software Foundation address in license notices\n t/lib-gpg: fix gpgconf stderr redirect to /dev/null\n rebase: fix stderr redirect in apply_autostash()\n notes: send \"Automatic notes merge failed\" messages to stderr\n branch doc: remove --set-upstream from synopsis\n completion: add '--copy' option to 'git branch'\n RelNotes: minor typo fixes in 2.15.1 draft\n t/lib-git-svn: cleanup inconsistent tab/space usage\n t/lib-git-svn.sh: improve svnserve tests with parallel make test\n RelNotes: minor typo fixes in 2.16.0 draft\n\nTorsten Bögershausen (3):\n add: introduce \"--renormalize\"\n convert: tighten the safe autocrlf handling\n t0027: Adapt the new MIX tests to Windows\n\nW. Trevor King (1):\n pull: pass --signoff/--no-signoff to \"git merge\"\n\nWei Shuyu (1):\n http: support CURLPROXY_HTTPS\n\nbrian m. carlson (31):\n walker: convert to struct object_id\n refs/files-backend: convert struct ref_to_prune to object_id\n refs: convert delete_ref and refs_delete_ref to struct object_id\n refs: convert update_ref and refs_update_ref to use struct object_id\n refs: prevent accidental NULL dereference in write_pseudoref\n refs: update ref transactions to use struct object_id\n Convert check_connected to use struct object_id\n refs: convert resolve_refdup and refs_resolve_refdup to struct object_id\n refs: convert read_ref and read_ref_full to object_id\n refs: convert dwim_ref and expand_ref to struct object_id\n builtin/reflog: convert remaining unsigned char uses to object_id\n refs: convert dwim_log to struct object_id\n pack-bitmap: convert traverse_bitmap_commit_list to object_id\n builtin/pack-objects: convert to struct object_id\n refs: convert peel_ref to struct object_id\n refs: convert read_ref_at to struct object_id\n refs: convert reflog_expire parameter to struct object_id\n sha1_file: convert index_path and index_fd to struct object_id\n Convert remaining callers of resolve_gitlink_ref to object_id\n refs: convert resolve_gitlink_ref to struct object_id\n worktree: convert struct worktree to object_id\n refs: convert resolve_ref_unsafe to struct object_id\n refs: convert peel_object to struct object_id\n refs: convert read_raw_ref backends to struct object_id\n refs/files-backend: convert static functions to object_id\n Documentation: enable compat-mode for Asciidoctor\n setup: expose enumerated repo info\n Add structure representing hash algorithm\n Integrate hash algorithm support with repo setup\n Switch empty tree and blob lookups to use hash abstraction\n Documentation: convert SubmittingPatches to AsciiDoc\n\nÆvar Arnfjörð Bjarmason (2):\n test-lib: add LIBPCRE1 & LIBPCRE2 prerequisites\n grep: fix segfault under -P + PCRE2 <=10.30 + (*NO_JIT)\n\nŁukasz Stelmach (1):\n git-gui: prevent double UTF-8 conversion\n\n"},{"id":"335519","messageId":"52672e8e-7ffe-a520-5f50-9e7fd15f28c5@gmail.com","threadId":"47507","inReplyTo":"xmqqfu7ui2af.fsf@gitster.mtv.corp.google.com","subject":"Re: [ANNOUNCE] Git v2.16.0-rc0","fromName":"Kaartic Sivaraam","fromEmail":"kaartic.sivaraam@gmail.com","sentAt":"2017-12-29T07:17:23Z","receivedAt":"2017-12-29T07:17:51Z","isPatch":false,"sender":{"key":"kaartic.sivaraam@gmail.com","avatar":"https://avatars.githubusercontent.com/u/12448084?v=4"},"body":"On Friday 29 December 2017 10:00 AM, Junio C Hamano wrote:\n> * \"git branch\" and \"git checkout -b\" are now forbidden from creating\n> a branch whose name is \"HEAD\".\n\n\"git branch\" already forbid a branch named \"HEAD\", didn't it? I thought \nwe just made \"git checkout -b\" to reject \"HEAD\" as a valid branch name, \nrecently.\n\n-- \nKaartic\n"},{"id":"335541","messageId":"1514567611.3270.28.camel@mad-scientist.net","threadId":"47507","inReplyTo":"xmqqfu7ui2af.fsf@gitster.mtv.corp.google.com","subject":"Re: [ANNOUNCE] Git v2.16.0-rc0","fromName":"Paul Smith","fromEmail":"paul@mad-scientist.net","sentAt":"2017-12-29T17:13:31Z","receivedAt":"2017-12-29T17:13:43Z","isPatch":false,"sender":{"key":"paul@mad-scientist.net","avatar":"https://avatars.githubusercontent.com/u/109636?v=4"},"body":"On Thu, 2017-12-28 at 20:30 -0800, Junio C Hamano wrote:\n> * The way \"git worktree add\" determines what branch to create from\n> where and checkout in the new worktree has been updated a bit.\n\nDoes this include the enhancements published a few weeks ago to allow\nworktrees to be created directly from remote branches without first\nchecking out the branch locally? I'm really looking forward to that\nchange...\n\nThanks!\n"},{"id":"335697","messageId":"CAGyf7-FQp4q2vvH1ponQvmVDTu0hiMSK1JKytQZ4O1i0MCnz7g@mail.gmail.com","threadId":"47507","inReplyTo":"xmqqfu7ui2af.fsf@gitster.mtv.corp.google.com","subject":"Re: [ANNOUNCE] Git v2.16.0-rc0","fromName":"Bryan Turner","fromEmail":"bturner@atlassian.com","sentAt":"2018-01-03T03:34:04Z","receivedAt":"2018-01-03T03:34:39Z","isPatch":false,"sender":{"key":"bturner@atlassian.com","avatar":"https://gravatar.com/avatar/16bcf3167981c1ef7c804e502642366d888a35b0d0b0a4ca01fdc442aa1acb1e?d=mp&s=160"},"body":"On Thu, Dec 28, 2017 at 8:30 PM, Junio C Hamano wrote:\n> An early preview release Git v2.16.0-rc0 is now available for\n> testing at the usual places. It is comprised of 435 non-merge\n> commits since v2.15.0, contributed by 76 people, 22 of which are\n> new faces.\n\n> Brandon Williams (24):\n> ssh: introduce a 'simple' ssh variant\n\n> Jonathan Nieder (10):\n> ssh test: make copy_ssh_wrapper_as clean up after itself\n> connect: move no_fork fallback to git_tcp_connect\n> connect: split git:// setup into a separate function\n> connect: split ssh command line options into separate function\n> connect: split ssh option computation to its own function\n> ssh: 'auto' variant to select between 'ssh' and 'simple'\n> ssh: 'simple' variant does not support -4/-6\n> ssh: 'simple' variant does not support --port\n> connect: correct style of C-style comment\n> generate-cmdlist: avoid non-deterministic output\n\nSorry for being late to the party on the \"simple\" variant for SSH, but\nwe've been doing some testing with 2.16.0-rc0 and noticed an\nunexpected issue.\n\nOur test environment is still on Ubuntu 12.04 LTS (it's a long story,\nbut one I doubt is unique to us), which means it's using OpenSSH 5.9.\nssh -G was added in OpenSSH 6.8 [1], circa March 2015, which means the\n\"auto\" detection \"fails\" and chooses \"simple\" instead of \"ssh\". But\nOpenSSH 5.9 _does_ support -4, -6 and -p. As a result, commands which\nhave been working without issue on all previous versions of Git start\nto fail saying:\n\ngit -c gc.auto=0 -c credential.helper= fetch --force --prune\n--progress ssh://localhost:64281/repo.git +refs/*:refs/*' exited with\ncode 128 saying: fatal: ssh variant 'simple' does not support setting\nport\n\nI know Ubuntu 12.04 LTS is end-of-life, but 14.04 LTS, which is\nrunning OpenSSH 6.6 [2], has the same issue. The following is from a\nfully patched 14.04.5:\n\nbturner@ubuntu:~$ cat /etc/*ease | head -4\nDISTRIB_ID=Ubuntu\nDISTRIB_RELEASE=14.04\nDISTRIB_CODENAME=trusty\nDISTRIB_DESCRIPTION=\"Ubuntu 14.04.5 LTS\"\n\nbturner@ubuntu:~$ ssh -V\nOpenSSH_6.6.1p1 Ubuntu-2ubuntu2.8, OpenSSL 1.0.1f 6 Jan 2014\n\nbturner@ubuntu:~$ ssh -G -p 7999 localhost\nunknown option -- G\nusage: ssh [-1246AaCfgKkMNnqsTtVvXxYy] [-b bind_address] [-c cipher_spec]\n [-D [bind_address:]port] [-E log_file] [-e escape_char]\n [-F configfile] [-I pkcs11] [-i identity_file]\n [-L [bind_address:]port:host:hostport] [-l login_name] [-m mac_spec]\n [-O ctl_cmd] [-o option] [-p port]\n [-Q cipher | cipher-auth | mac | kex | key]\n [-R [bind_address:]port:host:hostport] [-S ctl_path] [-W host:port]\n [-w local_tun[:remote_tun]] [user@]hostname [command]\n\nIs it possible to adjust the check, somehow, so it doesn't impact\nolder OpenSSH versions like this? As it stands, it seems likely a fair\nnumber of users who have an SSH command that does support -4, -6 and\n-p are going to end up getting \"penalized\" because it doesn't also\nsupport -G, and have to manually set their SSH variant to \"ssh\" (or\nsomething other than \"auto\") to avoid the automatic detection.\n\nI'd love to say I have a brilliant idea for how to work around this,\noh and here's a patch, but I don't. One option might be trying to\nactually review the output, and another might be to run \"ssh -V\", but\nboth of those have their own flaws (and the extra process forks aren't\n\"free\").\n\n[1] https://www.openssh.com/txt/release-6.8\n[2] https://launchpad.net/ubuntu/+source/openssh\n\nBest regards,\nBryan Turner\n"},{"id":"335699","messageId":"20180103050730.GA87855@aiede.mtv.corp.google.com","threadId":"47507","inReplyTo":"CAGyf7-FQp4q2vvH1ponQvmVDTu0hiMSK1JKytQZ4O1i0MCnz7g@mail.gmail.com","subject":"Re: [ANNOUNCE] Git v2.16.0-rc0","fromName":"Jonathan Nieder","fromEmail":"jrnieder@gmail.com","sentAt":"2018-01-03T05:07:30Z","receivedAt":"2018-01-03T05:07:56Z","isPatch":false,"sender":{"key":"jrnieder@gmail.com","avatar":"https://avatars.githubusercontent.com/u/281595?v=4"},"body":"Hi Bryan,\n\nBryan Turner wrote:\n\n> Our test environment is still on Ubuntu 12.04 LTS (it's a long story,\n> but one I doubt is unique to us), which means it's using OpenSSH 5.9.\n> ssh -G was added in OpenSSH 6.8 [1], circa March 2015, which means the\n> \"auto\" detection \"fails\" and chooses \"simple\" instead of \"ssh\". But\n> OpenSSH 5.9 _does_ support -4, -6 and -p. As a result, commands which\n> have been working without issue on all previous versions of Git start\n> to fail saying\n>\n> git -c gc.auto=0 -c credential.helper= fetch --force --prune --progress ssh://localhost:64281/repo.git +refs/*:refs/*' exited with code 128 saying: fatal: ssh variant 'simple' does not support setting port\n\nHm, that's not expected. git-config(1) says:\n\n\tBy default, Git determines the command line arguments to use\n\tbased on the basename of the configured SSH command\n\t(configured using the environment variable GIT_SSH or\n\tGIT_SSH_COMMAND or the config setting core.sshCommand). If the\n\tbasename is unrecognized, Git will attempt to detect support\n\tof OpenSSH options by [...]\n\nSo my first question is why the basename detection is not working for\nyou. What value of GIT_SSH, GIT_SSH_COMMAND, or core.sshCommand are\nyou using?\n\n> I know Ubuntu 12.04 LTS is end-of-life, but 14.04 LTS, which is\n> running OpenSSH 6.6 [2], has the same issue. The following is from a\n> fully patched 14.04.5:\n>\n> bturner@ubuntu:~$ cat /etc/*ease | head -4\n> DISTRIB_ID=Ubuntu\n> DISTRIB_RELEASE=14.04\n> DISTRIB_CODENAME=trusty\n> DISTRIB_DESCRIPTION=\"Ubuntu 14.04.5 LTS\"\n>\n> bturner@ubuntu:~$ ssh -V\n> OpenSSH_6.6.1p1 Ubuntu-2ubuntu2.8, OpenSSL 1.0.1f 6 Jan 2014\n>\n> bturner@ubuntu:~$ ssh -G -p 7999 localhost\n> unknown option -- G\n\nIt's good you caught this flaw in the detection. Would something like\nthe following make sense? If so, I can resend with a commit message\nand tests tomorrow or the day after.\n\ndiff --git i/Documentation/config.txt w/Documentation/config.txt\nindex 64c1dbba94..75eafd8db6 100644\n--- i/Documentation/config.txt\n+++ w/Documentation/config.txt\n@@ -2118,8 +2118,8 @@ ssh.variant::\n \tunrecognized, Git will attempt to detect support of OpenSSH\n \toptions by first invoking the configured SSH command with the\n \t`-G` (print configuration) option and will subsequently use\n-\tOpenSSH options (if that is successful) or no options besides\n-\tthe host and remote command (if it fails).\n+\tOpenSSH options if that is successful or a conservative set of\n+\tOpenSSH-style options if it fails.\n +\n The config variable `ssh.variant` can be set to override this detection.\n Valid values are `ssh` (to use OpenSSH options), `plink`, `putty`,\ndiff --git i/connect.c w/connect.c\nindex c3a014c5ba..3784c2be53 100644\n--- i/connect.c\n+++ w/connect.c\n@@ -941,10 +941,9 @@ static void push_ssh_options(struct argv_array *args, struct argv_array *env,\n \n \tif (flags & CONNECT_IPV4) {\n \t\tswitch (variant) {\n-\t\tcase VARIANT_AUTO:\n-\t\t\tBUG(\"VARIANT_AUTO passed to push_ssh_options\");\n \t\tcase VARIANT_SIMPLE:\n \t\t\tdie(\"ssh variant 'simple' does not support -4\");\n+\t\tcase VARIANT_AUTO:\n \t\tcase VARIANT_SSH:\n \t\tcase VARIANT_PLINK:\n \t\tcase VARIANT_PUTTY:\n@@ -953,10 +952,9 @@ static void push_ssh_options(struct argv_array *args, struct argv_array *env,\n \t\t}\n \t} else if (flags & CONNECT_IPV6) {\n \t\tswitch (variant) {\n-\t\tcase VARIANT_AUTO:\n-\t\t\tBUG(\"VARIANT_AUTO passed to push_ssh_options\");\n \t\tcase VARIANT_SIMPLE:\n \t\t\tdie(\"ssh variant 'simple' does not support -6\");\n+\t\tcase VARIANT_AUTO:\n \t\tcase VARIANT_SSH:\n \t\tcase VARIANT_PLINK:\n \t\tcase VARIANT_PUTTY:\n@@ -970,10 +968,9 @@ static void push_ssh_options(struct argv_array *args, struct argv_array *env,\n \n \tif (port) {\n \t\tswitch (variant) {\n-\t\tcase VARIANT_AUTO:\n-\t\t\tBUG(\"VARIANT_AUTO passed to push_ssh_options\");\n \t\tcase VARIANT_SIMPLE:\n \t\t\tdie(\"ssh variant 'simple' does not support setting port\");\n+\t\tcase VARIANT_AUTO:\n \t\tcase VARIANT_SSH:\n \t\t\targv_array_push(args, \"-p\");\n \t\t\tbreak;\n@@ -1026,7 +1023,7 @@ static void fill_ssh_args(struct child_process *conn, const char *ssh_host,\n \t\t\t\t VARIANT_SSH, port, flags);\n \t\targv_array_push(&detect.args, ssh_host);\n \n-\t\tvariant = run_command(&detect) ? VARIANT_SIMPLE : VARIANT_SSH;\n+\t\tvariant = run_command(&detect) ? VARIANT_AUTO : VARIANT_SSH;\n \t}\n \n \targv_array_push(&conn->args, ssh);\ndiff --git i/t/t5601-clone.sh w/t/t5601-clone.sh\nindex 0f895478f0..0224edc85b 100755\n--- i/t/t5601-clone.sh\n+++ w/t/t5601-clone.sh\n@@ -365,6 +365,11 @@ test_expect_success 'OpenSSH variant passes -4' '\n \texpect_ssh \"-4 -p 123\" myhost src\n '\n \n+test_expect_success 'OpenSSH passes GIT_PROTOCOL envvar' '\n+\tgit -c protocol.version=1 clone [myhost:123]:src ssh-v1-clone &&\n+\texpect_ssh \"-o SendEnv=GIT_PROTOCOL -p 123\" myhost src\n+'\n+\n test_expect_success 'variant can be overridden' '\n \tcopy_ssh_wrapper_as \"$TRASH_DIRECTORY/putty\" &&\n \tgit -c ssh.variant=putty clone -4 \"[myhost:123]:src\" ssh-putty-clone &&\n@@ -377,19 +382,32 @@ test_expect_success 'variant=auto picks based on basename' '\n \texpect_ssh \"-4 -P 123\" myhost src\n '\n \n-test_expect_success 'simple does not support -4/-6' '\n+test_expect_success 'variant=simple does not support -4/-6' '\n \tcopy_ssh_wrapper_as \"$TRASH_DIRECTORY/simple\" &&\n-\ttest_must_fail git clone -4 \"myhost:src\" ssh-4-clone-simple\n+\ttest_must_fail git -c ssh.variant=simple clone -4 \"myhost:src\" ssh-4-clone-simple\n '\n \n-test_expect_success 'simple does not support port' '\n+test_expect_success 'variant=simple does not support port' '\n \tcopy_ssh_wrapper_as \"$TRASH_DIRECTORY/simple\" &&\n-\ttest_must_fail git clone \"[myhost:123]:src\" ssh-bracket-clone-simple\n+\ttest_must_fail git -c ssh.variant=simple clone \"[myhost:123]:src\" ssh-bracket-clone-simple\n+'\n+\n+test_expect_success 'old OpenSSH passes -4 and -p' '\n+\tcopy_ssh_wrapper_as \"$TRASH_DIRECTORY/old\" &&\n+\tgit -c ssh.variant=auto clone -4 \"[myhost:123]:src\" old-ssh-clone &&\n+\texpect_ssh \"-4 -p 123\" myhost src\n '\n \n-test_expect_success 'uplink is treated as simple' '\n+test_expect_success 'old OpenSSH does not pass GIT_PROTOCOL envvar' '\n+\tcopy_ssh_wrapper_as \"$TRASH_DIRECTORY/old\" &&\n+\tgit -c protocol.version=1 -c ssh.variant=auto clone \"[myhost:123]:src\" old-ssh-protocol-v1 &&\n+\texpect_ssh \"-p 123\" myhost src\n+'\n+\n+test_expect_success 'uplink is treated as old OpenSSH' '\n \tcopy_ssh_wrapper_as \"$TRASH_DIRECTORY/uplink\" &&\n-\ttest_must_fail git clone \"[myhost:123]:src\" ssh-bracket-clone-uplink &&\n+\tgit -c protocol.version=1 clone \"[myhost:123]:src\" ssh-bracket-clone-uplink &&\n+\texpect_ssh \"-p 123\" myhost src &&\n \tgit clone \"myhost:src\" ssh-clone-uplink &&\n \texpect_ssh myhost src\n '\n@@ -405,8 +423,8 @@ test_expect_success 'OpenSSH-like uplink is treated as ssh' '\n \ttest_when_finished \"rm -f \\\"\\$TRASH_DIRECTORY/uplink\\\"\" &&\n \tGIT_SSH=\"$TRASH_DIRECTORY/uplink\" &&\n \ttest_when_finished \"GIT_SSH=\\\"\\$TRASH_DIRECTORY/ssh\\$X\\\"\" &&\n-\tgit clone \"[myhost:123]:src\" ssh-bracket-clone-sshlike-uplink &&\n-\texpect_ssh \"-p 123\" myhost src\n+\tgit -c protocol.version=1 clone \"[myhost:123]:src\" ssh-bracket-clone-sshlike-uplink &&\n+\texpect_ssh \"-o SendEnv=GIT_PROTOCOL -p 123\" myhost src\n '\n \n test_expect_success 'plink is treated specially (as putty)' '\n"},{"id":"335700","messageId":"20180103053516.GB87855@aiede.mtv.corp.google.com","threadId":"47507","inReplyTo":"CAGyf7-FQp4q2vvH1ponQvmVDTu0hiMSK1JKytQZ4O1i0MCnz7g@mail.gmail.com","subject":"Re: [ANNOUNCE] Git v2.16.0-rc0","fromName":"Jonathan Nieder","fromEmail":"jrnieder@gmail.com","sentAt":"2018-01-03T05:35:16Z","receivedAt":"2018-01-03T05:35:38Z","isPatch":false,"sender":{"key":"jrnieder@gmail.com","avatar":"https://avatars.githubusercontent.com/u/281595?v=4"},"body":"Hi,\n\nA few more notes.\n\nBryan Turner wrote:\n\n> bturner@ubuntu:~$ ssh -V\n> OpenSSH_6.6.1p1 Ubuntu-2ubuntu2.8, OpenSSL 1.0.1f 6 Jan 2014\n>\n> bturner@ubuntu:~$ ssh -G -p 7999 localhost\n> unknown option -- G\n> usage: ssh [-1246AaCfgKkMNnqsTtVvXxYy] [-b bind_address] [-c cipher_spec]\n[...]\n> Is it possible to adjust the check, somehow, so it doesn't impact\n> older OpenSSH versions like this? As it stands, it seems likely a fair\n> number of users who have an SSH command that does support -4, -6 and\n> -p are going to end up getting \"penalized\" because it doesn't also\n> support -G, and have to manually set their SSH variant to \"ssh\" (or\n> something other than \"auto\") to avoid the automatic detection.\n>\n> I'd love to say I have a brilliant idea for how to work around this,\n> oh and here's a patch, but I don't. One option might be trying to\n> actually review the output, and another might be to run \"ssh -V\", but\n> both of those have their own flaws (and the extra process forks aren't\n> \"free\").\n\nI have tomorrow off, so I've filed https://crbug.com/git/7 to make\nsure I remember to follow up the day after. Of course I'll be happy\nif someone updates that bug saying they've fixed it in the meantime.\n\nOne possibility would be to use -V as a fallback when -G fails, or\neven as a replacement for this usage of -G. To avoid misdetecting\nPuTTY and other ssh variants that also implement -V as OpenSSH, we\nwould have to parse the output. This would also misdetect a script\nthat does\n\n\thost=$1; shift\n\tssh \"$host\" -- \"$@\"\n\nas supporting OpenSSH options, when the use of -- ensures it doesn't.\n\nAnother possibility is to parse the output when -G fails. That's\nhacky, but I think it would work well! We would not have to be too\nclever, since we can look for the exact output produced by the\nversions of OpenSSH that we care about. This still has issues with\nscripts that forward arguments to OpenSSH, but at least those issues\nwould go away once the user updates their copy of ssh. ;-)\n\nAnother possibility is to pass options *before* -V:\n\n\tssh -p 7999 -V\n\nSince OpenSSH parses its arguments left-to-right, this gives similar\ninformation to what we did with -G, and scripts like\n\n\thost=$1; shift\n\tssh \"$host\" -- \"$@\"\n\nwould even be correctly detected as not supporting OpenSSH options.\nWe still would need to parse the output to distinguish OpenSSH from\nother ssh implementations like putty (unlike OpenSSH, putty saves up\nargument errors in an 'error' variable and forgets about them once it\nsees -V).\n\nTrying -G and falling back to -V seems like the simplest detection\nmechanism to me at the moment. I'm hoping I'm missing something\nsimple (another ssh option?) that allows avoiding this mess.\n\nRegardless, I think we should do something like [1] first to get rid\nof the regression. Thanks again for reporting it.\n\nSincerely,\nJonathan\n\n[1] https://public-inbox.org/git/20180103050730.GA87855@aiede.mtv.corp.google.com/\n"},{"id":"335702","messageId":"CAGyf7-EeBHwg50+MnUGF6=aE91TJY_iYSrg5gF1rp8ApsTsrvw@mail.gmail.com","threadId":"47507","inReplyTo":"20180103050730.GA87855@aiede.mtv.corp.google.com","subject":"Re: [ANNOUNCE] Git v2.16.0-rc0","fromName":"Bryan Turner","fromEmail":"bturner@atlassian.com","sentAt":"2018-01-03T05:41:43Z","receivedAt":"2018-01-03T05:42:01Z","isPatch":false,"sender":{"key":"bturner@atlassian.com","avatar":"https://gravatar.com/avatar/16bcf3167981c1ef7c804e502642366d888a35b0d0b0a4ca01fdc442aa1acb1e?d=mp&s=160"},"body":"On Tue, Jan 2, 2018 at 9:07 PM, Jonathan Nieder wrote:\n> Hi Bryan,\n>\n> Bryan Turner wrote:\n>\n>> Our test environment is still on Ubuntu 12.04 LTS (it's a long story,\n>> but one I doubt is unique to us), which means it's using OpenSSH 5.9.\n>> ssh -G was added in OpenSSH 6.8 [1], circa March 2015, which means the\n>> \"auto\" detection \"fails\" and chooses \"simple\" instead of \"ssh\". But\n>> OpenSSH 5.9 _does_ support -4, -6 and -p. As a result, commands which\n>> have been working without issue on all previous versions of Git start\n>> to fail saying\n>>\n>> git -c gc.auto=0 -c credential.helper= fetch --force --prune --progress ssh://localhost:64281/repo.git +refs/*:refs/*' exited with code 128 saying: fatal: ssh variant 'simple' does not support setting port\n>\n> Hm, that's not expected. git-config(1) says:\n>\n> By default, Git determines the command line arguments to use\n> based on the basename of the configured SSH command\n> (configured using the environment variable GIT_SSH or\n> GIT_SSH_COMMAND or the config setting core.sshCommand). If the\n> basename is unrecognized, Git will attempt to detect support\n> of OpenSSH options by [...]\n>\n> So my first question is why the basename detection is not working for\n> you. What value of GIT_SSH, GIT_SSH_COMMAND, or core.sshCommand are\n> you using?\n\nSo I'd been digging further into this for the last hour because I\nwasn't seeing quite the behavior I was expecting when I ran Git from\nthe command line on Ubuntu 12.04 or 14.04, and this nudged me to the\nright answer: We're setting GIT_SSH to a wrapper script. In our case,\nthat wrapper script is just calling OpenSSH's ssh with all the\nprovided arguments (plus a couple extra ones), but because we're\nsetting GIT_SSH at all, that's why the auto variant code is running.\nThat being the case, explicitly setting GIT_SSH_VARIANT=ssh may be the\ncorrect thing to do, to tell Git that we want to be treated like\n\"normal\" OpenSSH, as opposed to expecting Git to assume we behave like\nOpenSSH (when the Android repo use case clearly shows that assumption\nalso doesn't hold).\n\nBased on that, I'm not sure if you _actually_ need to change anything,\nand I apologize for not turning up that we were setting GIT_SSH before\nI bothered you.\n\nThanks immensely for looking into it, Jonathan, and my apologies again\nfor wasting your time!\nBryan Turner\n"},{"id":"335703","messageId":"20180103055049.GC87855@aiede.mtv.corp.google.com","threadId":"47507","inReplyTo":"CAGyf7-EeBHwg50+MnUGF6=aE91TJY_iYSrg5gF1rp8ApsTsrvw@mail.gmail.com","subject":"Re: [ANNOUNCE] Git v2.16.0-rc0","fromName":"Jonathan Nieder","fromEmail":"jrnieder@gmail.com","sentAt":"2018-01-03T05:50:49Z","receivedAt":"2018-01-03T05:51:09Z","isPatch":false,"sender":{"key":"jrnieder@gmail.com","avatar":"https://avatars.githubusercontent.com/u/281595?v=4"},"body":"Bryan Turner wrote:\n> On Tue, Jan 2, 2018 at 9:07 PM, Jonathan Nieder wrote:\n\n>> So my first question is why the basename detection is not working for\n>> you. What value of GIT_SSH, GIT_SSH_COMMAND, or core.sshCommand are\n>> you using?\n>\n> So I'd been digging further into this for the last hour because I\n> wasn't seeing quite the behavior I was expecting when I ran Git from\n> the command line on Ubuntu 12.04 or 14.04, and this nudged me to the\n> right answer: We're setting GIT_SSH to a wrapper script. In our case,\n> that wrapper script is just calling OpenSSH's ssh with all the\n> provided arguments (plus a couple extra ones), but because we're\n> setting GIT_SSH at all, that's why the auto variant code is running.\n> That being the case, explicitly setting GIT_SSH_VARIANT=ssh may be the\n> correct thing to do, to tell Git that we want to be treated like\n> \"normal\" OpenSSH, as opposed to expecting Git to assume we behave like\n> OpenSSH (when the Android repo use case clearly shows that assumption\n> also doesn't hold).\n\nAh, that's a comfort. Setting GIT_SSH_VARIANT would avoid this\nautodetection code and is the recommended thing to do.\n\nThat said, we can't go back in time and update everyone's tools to do\nthat (e.g. there is not even a release of repo with [1] out yet), so\nthis is still considered a regression and I'm glad you found it.\n\nJonathan\n\n[1] https://gerrit-review.googlesource.com/c/git-repo/+/134950\n"},{"id":"335785","messageId":"xmqqk1wyhcey.fsf@gitster.mtv.corp.google.com","threadId":"47507","inReplyTo":"20180103050730.GA87855@aiede.mtv.corp.google.com","subject":"Re: [ANNOUNCE] Git v2.16.0-rc0","fromName":"Junio C Hamano","fromEmail":"gitster@pobox.com","sentAt":"2018-01-03T21:15:17Z","receivedAt":"2018-01-03T21:15:26Z","isPatch":false,"sender":{"key":"gitster@pobox.com","avatar":"https://avatars.githubusercontent.com/u/54884?v=4"},"body":"Jonathan Nieder writes:\n\n> It's good you caught this flaw in the detection. Would something like\n> the following make sense? If so, I can resend with a commit message\n> and tests tomorrow or the day after.\n\nSo the idea is to keep the 'simple' for implementations that do not\nsupport OpenSSH options, but declare that the auto-detection was\noverly conservative and assume that -4/6/p are supported by\neverybody?\n\nThis change means that those who were meant to be helped by the\noriginal change, that introduced 'simple' and made the (overly\nconservative) auto-detection, would now have to explicitly set\nssh.variant to simple, because otherwise they will be passed one of\nthese three options. Am I reading the intention of the change\ncorrectly? If so, I tend to agree that it is lesser of the two\nevils to make sure things continue to work for older openssh users\nwith their current setting like this patch does, even with the cost\nof telling users with implementations that do not honor -4/6/p to\nset things up manually, I guess.\n\nThanks, both, for digging this issue through.\n\n>\n> diff --git i/Documentation/config.txt w/Documentation/config.txt\n> index 64c1dbba94..75eafd8db6 100644\n> --- i/Documentation/config.txt\n> +++ w/Documentation/config.txt\n> @@ -2118,8 +2118,8 @@ ssh.variant::\n> \tunrecognized, Git will attempt to detect support of OpenSSH\n> \toptions by first invoking the configured SSH command with the\n> \t`-G` (print configuration) option and will subsequently use\n> -\tOpenSSH options (if that is successful) or no options besides\n> -\tthe host and remote command (if it fails).\n> +\tOpenSSH options if that is successful or a conservative set of\n> +\tOpenSSH-style options if it fails.\n> +\n> The config variable `ssh.variant` can be set to override this detection.\n> Valid values are `ssh` (to use OpenSSH options), `plink`, `putty`,\n> diff --git i/connect.c w/connect.c\n> index c3a014c5ba..3784c2be53 100644\n> --- i/connect.c\n> +++ w/connect.c\n> @@ -941,10 +941,9 @@ static void push_ssh_options(struct argv_array *args, struct argv_array *env,\n> \n> \tif (flags & CONNECT_IPV4) {\n> \t\tswitch (variant) {\n> -\t\tcase VARIANT_AUTO:\n> -\t\t\tBUG(\"VARIANT_AUTO passed to push_ssh_options\");\n> \t\tcase VARIANT_SIMPLE:\n> \t\t\tdie(\"ssh variant 'simple' does not support -4\");\n> +\t\tcase VARIANT_AUTO:\n> \t\tcase VARIANT_SSH:\n> \t\tcase VARIANT_PLINK:\n> \t\tcase VARIANT_PUTTY:\n> @@ -953,10 +952,9 @@ static void push_ssh_options(struct argv_array *args, struct argv_array *env,\n> \t\t}\n> \t} else if (flags & CONNECT_IPV6) {\n> \t\tswitch (variant) {\n> -\t\tcase VARIANT_AUTO:\n> -\t\t\tBUG(\"VARIANT_AUTO passed to push_ssh_options\");\n> \t\tcase VARIANT_SIMPLE:\n> \t\t\tdie(\"ssh variant 'simple' does not support -6\");\n> +\t\tcase VARIANT_AUTO:\n> \t\tcase VARIANT_SSH:\n> \t\tcase VARIANT_PLINK:\n> \t\tcase VARIANT_PUTTY:\n> @@ -970,10 +968,9 @@ static void push_ssh_options(struct argv_array *args, struct argv_array *env,\n> \n> \tif (port) {\n> \t\tswitch (variant) {\n> -\t\tcase VARIANT_AUTO:\n> -\t\t\tBUG(\"VARIANT_AUTO passed to push_ssh_options\");\n> \t\tcase VARIANT_SIMPLE:\n> \t\t\tdie(\"ssh variant 'simple' does not support setting port\");\n> +\t\tcase VARIANT_AUTO:\n> \t\tcase VARIANT_SSH:\n> \t\t\targv_array_push(args, \"-p\");\n> \t\t\tbreak;\n> @@ -1026,7 +1023,7 @@ static void fill_ssh_args(struct child_process *conn, const char *ssh_host,\n> \t\t\t\t VARIANT_SSH, port, flags);\n> \t\targv_array_push(&detect.args, ssh_host);\n> \n> -\t\tvariant = run_command(&detect) ? VARIANT_SIMPLE : VARIANT_SSH;\n> +\t\tvariant = run_command(&detect) ? VARIANT_AUTO : VARIANT_SSH;\n> \t}\n> \n> \targv_array_push(&conn->args, ssh);\n> diff --git i/t/t5601-clone.sh w/t/t5601-clone.sh\n> index 0f895478f0..0224edc85b 100755\n> --- i/t/t5601-clone.sh\n> +++ w/t/t5601-clone.sh\n> @@ -365,6 +365,11 @@ test_expect_success 'OpenSSH variant passes -4' '\n> \texpect_ssh \"-4 -p 123\" myhost src\n> '\n> \n> +test_expect_success 'OpenSSH passes GIT_PROTOCOL envvar' '\n> +\tgit -c protocol.version=1 clone [myhost:123]:src ssh-v1-clone &&\n> +\texpect_ssh \"-o SendEnv=GIT_PROTOCOL -p 123\" myhost src\n> +'\n> +\n> test_expect_success 'variant can be overridden' '\n> \tcopy_ssh_wrapper_as \"$TRASH_DIRECTORY/putty\" &&\n> \tgit -c ssh.variant=putty clone -4 \"[myhost:123]:src\" ssh-putty-clone &&\n> @@ -377,19 +382,32 @@ test_expect_success 'variant=auto picks based on basename' '\n> \texpect_ssh \"-4 -P 123\" myhost src\n> '\n> \n> -test_expect_success 'simple does not support -4/-6' '\n> +test_expect_success 'variant=simple does not support -4/-6' '\n> \tcopy_ssh_wrapper_as \"$TRASH_DIRECTORY/simple\" &&\n> -\ttest_must_fail git clone -4 \"myhost:src\" ssh-4-clone-simple\n> +\ttest_must_fail git -c ssh.variant=simple clone -4 \"myhost:src\" ssh-4-clone-simple\n> '\n> \n> -test_expect_success 'simple does not support port' '\n> +test_expect_success 'variant=simple does not support port' '\n> \tcopy_ssh_wrapper_as \"$TRASH_DIRECTORY/simple\" &&\n> -\ttest_must_fail git clone \"[myhost:123]:src\" ssh-bracket-clone-simple\n> +\ttest_must_fail git -c ssh.variant=simple clone \"[myhost:123]:src\" ssh-bracket-clone-simple\n> +'\n> +\n> +test_expect_success 'old OpenSSH passes -4 and -p' '\n> +\tcopy_ssh_wrapper_as \"$TRASH_DIRECTORY/old\" &&\n> +\tgit -c ssh.variant=auto clone -4 \"[myhost:123]:src\" old-ssh-clone &&\n> +\texpect_ssh \"-4 -p 123\" myhost src\n> '\n> \n> -test_expect_success 'uplink is treated as simple' '\n> +test_expect_success 'old OpenSSH does not pass GIT_PROTOCOL envvar' '\n> +\tcopy_ssh_wrapper_as \"$TRASH_DIRECTORY/old\" &&\n> +\tgit -c protocol.version=1 -c ssh.variant=auto clone \"[myhost:123]:src\" old-ssh-protocol-v1 &&\n> +\texpect_ssh \"-p 123\" myhost src\n> +'\n> +\n> +test_expect_success 'uplink is treated as old OpenSSH' '\n> \tcopy_ssh_wrapper_as \"$TRASH_DIRECTORY/uplink\" &&\n> -\ttest_must_fail git clone \"[myhost:123]:src\" ssh-bracket-clone-uplink &&\n> +\tgit -c protocol.version=1 clone \"[myhost:123]:src\" ssh-bracket-clone-uplink &&\n> +\texpect_ssh \"-p 123\" myhost src &&\n> \tgit clone \"myhost:src\" ssh-clone-uplink &&\n> \texpect_ssh myhost src\n> '\n> @@ -405,8 +423,8 @@ test_expect_success 'OpenSSH-like uplink is treated as ssh' '\n> \ttest_when_finished \"rm -f \\\"\\$TRASH_DIRECTORY/uplink\\\"\" &&\n> \tGIT_SSH=\"$TRASH_DIRECTORY/uplink\" &&\n> \ttest_when_finished \"GIT_SSH=\\\"\\$TRASH_DIRECTORY/ssh\\$X\\\"\" &&\n> -\tgit clone \"[myhost:123]:src\" ssh-bracket-clone-sshlike-uplink &&\n> -\texpect_ssh \"-p 123\" myhost src\n> +\tgit -c protocol.version=1 clone \"[myhost:123]:src\" ssh-bracket-clone-sshlike-uplink &&\n> +\texpect_ssh \"-o SendEnv=GIT_PROTOCOL -p 123\" myhost src\n> '\n> \n> test_expect_success 'plink is treated specially (as putty)' '\n"},{"id":"335860","messageId":"20180104201804.GE2641@hank","threadId":"47507","inReplyTo":"1514567611.3270.28.camel@mad-scientist.net","subject":"Re: [ANNOUNCE] Git v2.16.0-rc0","fromName":"Thomas Gummerer","fromEmail":"t.gummerer@gmail.com","sentAt":"2018-01-04T20:18:04Z","receivedAt":"2018-01-04T20:16:01Z","isPatch":false,"sender":{"key":"t.gummerer@gmail.com","avatar":"https://avatars.githubusercontent.com/u/191004?v=4"},"body":"On 12/29, Paul Smith wrote:\n> On Thu, 2017-12-28 at 20:30 -0800, Junio C Hamano wrote:\n> > * The way \"git worktree add\" determines what branch to create from\n> > where and checkout in the new worktree has been updated a bit.\n> \n> Does this include the enhancements published a few weeks ago to allow\n> worktrees to be created directly from remote branches without first\n> checking out the branch locally? I'm really looking forward to that\n> change...\n\nYes, this release will include that. It would be awesome if you could\ntest the rc, now is the best time to scream if something about it is\nnot as you'd expect :)\n\n> Thanks!\n"},{"id":"335862","messageId":"alpine.DEB.2.21.1.1801042132540.32@MININT-6BKU6QN.europe.corp.microsoft.com","threadId":"47507","inReplyTo":"xmqqfu7ui2af.fsf@gitster.mtv.corp.google.com","subject":"Re: [ANNOUNCE] Git v2.16.0-rc0","fromName":"Johannes Schindelin","fromEmail":"johannes.schindelin@gmx.de","sentAt":"2018-01-04T20:33:55Z","receivedAt":"2018-01-04T20:34:18Z","isPatch":false,"sender":{"key":"johannes.schindelin@gmx.de","avatar":"https://avatars.githubusercontent.com/u/127790?v=4"},"body":"Hi,\n\nOn Thu, 28 Dec 2017, Junio C Hamano wrote:\n\n> An early preview release Git v2.16.0-rc0 is now available for\n> testing at the usual places.\n\nAnd a corresponding Git for Windows prerelease is also available:\n\nhttps://github.com/git-for-windows/git/releases/tag/v2.16.0-rc0.windows.1\n\nCiao,\nJohannes\n"},{"id":"335869","messageId":"1515101753.2717.690.camel@mad-scientist.net","threadId":"47507","inReplyTo":"20180104201804.GE2641@hank","subject":"Re: [ANNOUNCE] Git v2.16.0-rc0","fromName":"Paul Smith","fromEmail":"paul@mad-scientist.net","sentAt":"2018-01-04T21:35:53Z","receivedAt":"2018-01-04T21:36:03Z","isPatch":false,"sender":{"key":"paul@mad-scientist.net","avatar":"https://avatars.githubusercontent.com/u/109636?v=4"},"body":"On Thu, 2018-01-04 at 20:18 +0000, Thomas Gummerer wrote:\n> On 12/29, Paul Smith wrote:\n> > On Thu, 2017-12-28 at 20:30 -0800, Junio C Hamano wrote:\n> > > * The way \"git worktree add\" determines what branch to create\n> > > from where and checkout in the new worktree has been updated a\n> > > bit.\n> > \n> > Does this include the enhancements published a few weeks ago to\n> > allow worktrees to be created directly from remote branches without\n> > first checking out the branch locally? I'm really looking forward\n> > to that change...\n> \n> Yes, this release will include that. It would be awesome if you\n> could test the rc, now is the best time to scream if something about\n> it is not as you'd expect :)\n\nOK, I pulled and built this locally, and indeed the straightforward\n\"git worktree add \" works just as I'd hoped! Nice!\n\nI'll play with it more and let you know if I hit issues.\n"},{"id":"349678","messageId":"20180608045028.GA17528@sigill.intra.peff.net","threadId":"47507","inReplyTo":"20180103053516.GB87855@aiede.mtv.corp.google.com","subject":"Re: [ANNOUNCE] Git v2.16.0-rc0","fromName":"Jeff King","fromEmail":"peff@peff.net","sentAt":"2018-06-08T04:50:28Z","receivedAt":"2018-06-08T04:50:33Z","isPatch":false,"sender":{"key":"peff@peff.net","avatar":"https://avatars.githubusercontent.com/u/45925?v=4"},"body":"On Tue, Jan 02, 2018 at 09:35:16PM -0800, Jonathan Nieder wrote:\n\n> > bturner@ubuntu:~$ ssh -V\n> > OpenSSH_6.6.1p1 Ubuntu-2ubuntu2.8, OpenSSL 1.0.1f 6 Jan 2014\n> >\n> > bturner@ubuntu:~$ ssh -G -p 7999 localhost\n> > unknown option -- G\n> > usage: ssh [-1246AaCfgKkMNnqsTtVvXxYy] [-b bind_address] [-c cipher_spec]\n> [...]\n> > Is it possible to adjust the check, somehow, so it doesn't impact\n> > older OpenSSH versions like this? As it stands, it seems likely a fair\n> > number of users who have an SSH command that does support -4, -6 and\n> > -p are going to end up getting \"penalized\" because it doesn't also\n> > support -G, and have to manually set their SSH variant to \"ssh\" (or\n> > something other than \"auto\") to avoid the automatic detection.\n> >\n> > I'd love to say I have a brilliant idea for how to work around this,\n> > oh and here's a patch, but I don't. One option might be trying to\n> > actually review the output, and another might be to run \"ssh -V\", but\n> > both of those have their own flaws (and the extra process forks aren't\n> > \"free\").\n> \n> I have tomorrow off, so I've filed https://crbug.com/git/7 to make\n> sure I remember to follow up the day after. Of course I'll be happy\n> if someone updates that bug saying they've fixed it in the meantime.\n\nIt doesn't look like we ever applied anything to deal with this\nregression. Just FYI, this bit me today when upgrading my git on a\nsystem that has an ssh wrapper that understands \"-p\" just fine, but not\n\"-G\". So the behavior described in [1], namely to just fallback to\nassuming some basic openssh-ish options, would have worked for me.\n\nTo be honest, I could easily see an argument that I _should_ be setting\nGIT_SSH_VARIANT to explain what my wrapper is expecting, even though it\nhappened to work before. But it seems like this discussion ended in\nfavor of calling this a regression that should be fixed, and AFAICT\nnothing happened after. So I thought I'd ping and mention one more data\npoint.\n\n-Peff\n\n[1] https://public-inbox.org/git/xmqqk1wyhcey.fsf@gitster.mtv.corp.google.com/\n"},{"id":"349989","messageId":"xmqqzi009deu.fsf@gitster-ct.c.googlers.com","threadId":"47507","inReplyTo":"20180608045028.GA17528@sigill.intra.peff.net","subject":"Re: [ANNOUNCE] Git v2.16.0-rc0","fromName":"Junio C Hamano","fromEmail":"gitster@pobox.com","sentAt":"2018-06-12T16:29:13Z","receivedAt":"2018-06-12T16:29:19Z","isPatch":false,"sender":{"key":"gitster@pobox.com","avatar":"https://avatars.githubusercontent.com/u/54884?v=4"},"body":"Jeff King writes:\n\n> To be honest, I could easily see an argument that I _should_ be setting\n> GIT_SSH_VARIANT to explain what my wrapper is expecting, even though it\n> happened to work before.\n\nThe way I read that message is that the patch proposed in\n\n https://public-inbox.org/git/20180103050730.GA87855@aiede.mtv.corp.google.com\n\nis \"lesser of two evils\" in that it is still evil because somebody\nstill has to be asked to explicitly set \"variant\" anyway but the\nchanging what 'simple' means in the middle would mean those who did\nnot have to set it now have to. So, you should be setting it, even\nif we adopt the patch, right? ;-)\n\n> But it seems like this discussion ended in\n> favor of calling this a regression that should be fixed, and AFAICT\n> nothing happened after. So I thought I'd ping and mention one more data\n> point.\n\nMy impression is that regression \"fix\" does not exist---rather, it\nwas a proposal (and it may make a better tradeoff than the status\nquo) to help users of older OpenSSH by inconveniencing users of\ndifferent implementations that do -4/6/p differently from OpenSSH.\n\nIn any case, from where I sit, I am still waiting for this offer\nby Jonathan\n\n> It's good you caught this flaw in the detection. Would something like\n> the following make sense? If so, I can resend with a commit message\n> and tests tomorrow or the day after.\n\nto be followed up ;-)\n\n"},{"id":"350183","messageId":"20180614183018.GA1911@sigill.intra.peff.net","threadId":"47507","inReplyTo":"xmqqzi009deu.fsf@gitster-ct.c.googlers.com","subject":"Re: [ANNOUNCE] Git v2.16.0-rc0","fromName":"Jeff King","fromEmail":"peff@peff.net","sentAt":"2018-06-14T18:30:19Z","receivedAt":"2018-06-14T18:30:25Z","isPatch":false,"sender":{"key":"peff@peff.net","avatar":"https://avatars.githubusercontent.com/u/45925?v=4"},"body":"On Tue, Jun 12, 2018 at 09:29:13AM -0700, Junio C Hamano wrote:\n\n> Jeff King writes:\n> \n> > To be honest, I could easily see an argument that I _should_ be setting\n> > GIT_SSH_VARIANT to explain what my wrapper is expecting, even though it\n> > happened to work before.\n> \n> The way I read that message is that the patch proposed in\n> \n> https://public-inbox.org/git/20180103050730.GA87855@aiede.mtv.corp.google.com\n> \n> is \"lesser of two evils\" in that it is still evil because somebody\n> still has to be asked to explicitly set \"variant\" anyway but the\n> changing what 'simple' means in the middle would mean those who did\n> not have to set it now have to. So, you should be setting it, even\n> if we adopt the patch, right? ;-)\n\nNo, my wrapper _isn't_ simple. It passes most options to openssh, but\njust doesn't understand the \"-G\" probing. So if the default was\nopenssh-like instead of \"simple\", then that would work fine without me\nsetting anything, just like it did before.\n\nWhich I thought was where the discussion ended up, but perhaps I'm\nmisunderstanding.\n\n> My impression is that regression \"fix\" does not exist---rather, it\n> was a proposal (and it may make a better tradeoff than the status\n> quo) to help users of older OpenSSH by inconveniencing users of\n> different implementations that do -4/6/p differently from OpenSSH.\n\nRight. That fixes the regression for openssh users, at the cost of not\nhelp people on other implementations automatically. But those people\nhave to set the flag anyway, since the current behavior is \"whoops, we\ndon't know how to support you, set the flag\".\n\nTo be clear, I've already fixed my setup, and it wasn't that big a deal.\nAnd I doubt all that many people would be affected either way. My use\ncase here was literally instrumenting the ssh calls of the git client as\npart of a regression test suite. How many git test suites could there\npossibly be? :)\n\nSo I'm OK if we just leave it as-is. It's mostly that I dug into the\nthread and was left with the impression that it was an unfinished\nleftover that we meant to do.\n\n> In any case, from where I sit, I am still waiting for this offer\n> by Jonathan\n> \n> > It's good you caught this flaw in the detection. Would something like\n> > the following make sense? If so, I can resend with a commit message\n> > and tests tomorrow or the day after.\n> \n> to be followed up ;-)\n\nYes, that was the part that left the impression. :)\n\n-Peff\n"},{"id":"350186","messageId":"20180614185522.GA255581@aiede.svl.corp.google.com","threadId":"47507","inReplyTo":"20180614183018.GA1911@sigill.intra.peff.net","subject":"Re: [ANNOUNCE] Git v2.16.0-rc0","fromName":"Jonathan Nieder","fromEmail":"jrnieder@gmail.com","sentAt":"2018-06-14T18:55:22Z","receivedAt":"2018-06-14T18:55:32Z","isPatch":false,"sender":{"key":"jrnieder@gmail.com","avatar":"https://avatars.githubusercontent.com/u/281595?v=4"},"body":"Hi,\n\nSorry for the slow replies. I was out of office earlier and am back\nnow.\n\nJeff King wrote:\n> On Tue, Jun 12, 2018 at 09:29:13AM -0700, Junio C Hamano wrote:\n>> Jeff King writes:\n\n>>> To be honest, I could easily see an argument that I _should_ be setting\n>>> GIT_SSH_VARIANT to explain what my wrapper is expecting, even though it\n>>> happened to work before.\n\nYes, I encourage anyone setting GIT_SSH to also set GIT_SSH_VARIANT.\nThis way, there can be no confusion about what the setter intends.\n\nThe autodetection is in the category of \"necessary evil\" that wouldn't\nexist if we were starting over.\n\n[...]\n> No, my wrapper _isn't_ simple. It passes most options to openssh, but\n> just doesn't understand the \"-G\" probing. So if the default was\n> openssh-like instead of \"simple\", then that would work fine without me\n> setting anything, just like it did before.\n>\n> Which I thought was where the discussion ended up, but perhaps I'm\n> misunderstanding.\n\nDo you mean that it doesn't pass \"-G\" through, or that when using old\nversions of openssh that doesn't support \"-G\" the probing fails?\n\nIf the former, then detecting the wrapper as something other than\n\"ssh\" is intended behavior (though we might want to change what that\nsomething is, as discussed in the previous thread). If the latter,\nthen this is https://crbug.com/git/7 which I consider to be a bug.\n\n[...]\n> So I'm OK if we just leave it as-is. It's mostly that I dug into the\n> thread and was left with the impression that it was an unfinished\n> leftover that we meant to do.\n[...]\n>> In any case, from where I sit, I am still waiting for this offer\n>> by Jonathan\n>>\n>>> It's good you caught this flaw in the detection. Would something like\n>>> the following make sense? If so, I can resend with a commit message\n>>> and tests tomorrow or the day after.\n>>\n>> to be followed up ;-)\n>\n> Yes, that was the part that left the impression. :)\n\nThanks for the poke.\n\nThe patch [1] didn't leave me super happy, since it means there is yet\nanother ssh variant for people to understand, in order to accomodate\nan old version of OpenSSH that is going to go away eventually. In\nDebian, modern versions of git declare an incompatibility with old\nversions of OpenSSH in their package metadata to avoid this issue.\n\nIn my defense, I said \"If so\" and no one seemed too enthusiastic about\nthe patch making sense. ;-)\n\nI'll take another look and probably resend.\n\nSincerely,\nJonathan\n\n[1] https://public-inbox.org/git/20180103050730.GA87855@aiede.mtv.corp.google.com/\n"},{"id":"350193","messageId":"20180614193943.GA2226@sigill.intra.peff.net","threadId":"47507","inReplyTo":"20180614185522.GA255581@aiede.svl.corp.google.com","subject":"Re: [ANNOUNCE] Git v2.16.0-rc0","fromName":"Jeff King","fromEmail":"peff@peff.net","sentAt":"2018-06-14T19:39:43Z","receivedAt":"2018-06-14T19:39:47Z","isPatch":false,"sender":{"key":"peff@peff.net","avatar":"https://avatars.githubusercontent.com/u/45925?v=4"},"body":"On Thu, Jun 14, 2018 at 11:55:22AM -0700, Jonathan Nieder wrote:\n\n> > No, my wrapper _isn't_ simple. It passes most options to openssh, but\n> > just doesn't understand the \"-G\" probing. So if the default was\n> > openssh-like instead of \"simple\", then that would work fine without me\n> > setting anything, just like it did before.\n> >\n> > Which I thought was where the discussion ended up, but perhaps I'm\n> > misunderstanding.\n> \n> Do you mean that it doesn't pass \"-G\" through, or that when using old\n> versions of openssh that doesn't support \"-G\" the probing fails?\n\nIt just doesn't pass \"-G\" through.\n\n> If the former, then detecting the wrapper as something other than\n> \"ssh\" is intended behavior (though we might want to change what that\n> something is, as discussed in the previous thread). If the latter,\n> then this is https://crbug.com/git/7 which I consider to be a bug.\n\nI certainly see the argument that \"well, if it doesn't do '-G' then it's\nnot _really_ openssh\". My counter to that is that we don't actually\n_care_ about -G (and never did before recently). It's just a proxy for\n\"do we understand -p\", which my script does understand. My wrapper might\neventually break if we depend on new options (like \"-o SendEnv\"), but\nthe worst case there is generally no different before or after your\npatch: the command barfs.\n\nI say \"generally\" because of course you can come up with an example\nwhere my script quietly interprets \"-o\" as something else, but it seems\nlike most uses there would cause an error. And anyway, by making me set\nGIT_SSH_VARIANT all we've bought is plausible deniability that it's _my_\nfault for doing so when my script doesn't handle the new option\ngracefully. ;)\n\nBut again, I'm just describing what makes sense to me. If you feel\nstrongly about requiring the variant to be explicitly specified, I can\ncertainly live with that.\n\n-Peff\n"},{"id":"350229","messageId":"20180615042033.GB255581@aiede.svl.corp.google.com","threadId":"47507","inReplyTo":"20180614193943.GA2226@sigill.intra.peff.net","subject":"Re: [ANNOUNCE] Git v2.16.0-rc0","fromName":"Jonathan Nieder","fromEmail":"jrnieder@gmail.com","sentAt":"2018-06-15T04:20:33Z","receivedAt":"2018-06-15T04:20:40Z","isPatch":false,"sender":{"key":"jrnieder@gmail.com","avatar":"https://avatars.githubusercontent.com/u/281595?v=4"},"body":"Hi,\n\nJeff King wrote:\n> On Thu, Jun 14, 2018 at 11:55:22AM -0700, Jonathan Nieder wrote:\n\n>> Do you mean that it doesn't pass \"-G\" through, or that when using old\n>> versions of openssh that doesn't support \"-G\" the probing fails?\n>\n> It just doesn't pass \"-G\" through.\n\nThanks.\n\n>> If the former, then detecting the wrapper as something other than\n>> \"ssh\" is intended behavior (though we might want to change what that\n>> something is, as discussed in the previous thread). If the latter,\n>> then this is https://crbug.com/git/7 which I consider to be a bug.\n>\n> I certainly see the argument that \"well, if it doesn't do '-G' then it's\n> not _really_ openssh\". My counter to that is that we don't actually\n> _care_ about -G (and never did before recently). It's just a proxy for\n> \"do we understand -p\", which my script does understand. My wrapper might\n> eventually break if we depend on new options (like \"-o SendEnv\")\n\nExactly: what we want to detect is not whether the script happens to\nsupport the OpenSSH options we use today, but whether it is a thin\nwrapper around OpenSSH that supports *all* options.\n\nThis wrapper definitely does not qualify. As we start to use more\nOpenSSH options, we are likely to break it. As you say,\n\n> the worst case there is generally no different before or after your\n> patch: the command barfs.\n\nbut as a caller, Git has no way to know that: it is easily possible\nthat the wrapper would ignore the new option we want to pass, for\nexample.\n\nIn other words, I think your response is based on the interface that\nwe previously, foolishly advertised: \"We will pass exactly these\noptions to your wrapper\". Except those options changed over time,\nmultiple times. It was a terrible interface.\n\nWhat we've switched to is a versioned interface. By setting\nGIT_SSH_VARIANT=simple, you are asking Git to promise to pass exactly\n options. If Git has a new option it wants to pass (like the \"-o\nSendEnv\" thing) but can live without it, then it can avoid breaking\nyour wrapper and continue to follow this new promise.\n\nThe trouble is that GIT_SSH_VARIANT=simple is too... simple. You\nwould like a variant that passes in [-p port] [-4] [-6] as well. We\ndidn't implement that because we didn't have the attention of any\nwrapper writer who wanted it; in absence of a potential user, we\ndecided to wait for a user to propose the interface they want. Now we\ncan celebrate, since that day has come.\n\nHow would you like your ssh variant to work? Some possibilities to\nget the thought process going:\n\n A. Would you want to set a variable 'GIT_SSH_SUPPORTS_OPTIONS=p46'\n to inform Git about what options you support?\n\n B. Alternatively, what about a 'GIT_SSH_VARIANT=capabilities' variant\n that calls \"your-ssh-variant --capabailities\" to get a\n machine-readable list of capabilities it supports?\n\n C. Alternatively, would you like all parameters to come in on stdin,\n credential helper style?\n\n D. Other ideas?\n\n[...]\n> But again, I'm just describing what makes sense to me. If you feel\n> strongly about requiring the variant to be explicitly specified, I can\n> certainly live with that.\n\nI think I do feel strongly.\n\nIf you were using an old version of OpenSSH, this would be a reason to\nrevive the old patch, but I'm tempted to stall longer just to get more\nuse cases like this to come out of the woodwork.\n\nOf course that's a dangerous thought process. Probably by tomorrow\nI'll think better of it and revive the patch. In the meantime I would\nbe happy to see your answers to the questions above.\n\nSincerely,\nJonathan\n"},{"id":"350232","messageId":"20180615051321.GA5840@sigill.intra.peff.net","threadId":"47507","inReplyTo":"20180615042033.GB255581@aiede.svl.corp.google.com","subject":"Re: [ANNOUNCE] Git v2.16.0-rc0","fromName":"Jeff King","fromEmail":"peff@peff.net","sentAt":"2018-06-15T05:13:21Z","receivedAt":"2018-06-15T05:13:27Z","isPatch":false,"sender":{"key":"peff@peff.net","avatar":"https://avatars.githubusercontent.com/u/45925?v=4"},"body":"On Thu, Jun 14, 2018 at 09:20:33PM -0700, Jonathan Nieder wrote:\n\n> What we've switched to is a versioned interface. By setting\n> GIT_SSH_VARIANT=simple, you are asking Git to promise to pass exactly\n> options. If Git has a new option it wants to pass (like the \"-o\n> SendEnv\" thing) but can live without it, then it can avoid breaking\n> your wrapper and continue to follow this new promise.\n> \n> The trouble is that GIT_SSH_VARIANT=simple is too... simple. You\n> would like a variant that passes in [-p port] [-4] [-6] as well. We\n> didn't implement that because we didn't have the attention of any\n> wrapper writer who wanted it; in absence of a potential user, we\n> decided to wait for a user to propose the interface they want. Now we\n> can celebrate, since that day has come.\n\nI'm not sure I'm celebrating. It seems like work for not much benefit.\nI'd just as soon use VARIANT=ssh and deal with any fallouts if my script\ndoes not behave exactly like openssh in all regards.\n\n> How would you like your ssh variant to work? Some possibilities to\n> get the thought process going:\n> \n> A. Would you want to set a variable 'GIT_SSH_SUPPORTS_OPTIONS=p46'\n> to inform Git about what options you support?\n\nNot really. That just creates more work when I have to later use \"op46\"\nto support \"-o\", even though my script already handles it fine (or\nworse, since \"-o\" is open-ended).\n\n> B. Alternatively, what about a 'GIT_SSH_VARIANT=capabilities' variant\n> that calls \"your-ssh-variant --capabailities\" to get a\n> machine-readable list of capabilities it supports?\n\nNot really. Now I have to implement this --capabilities thing. This is\nliterally a 10 line script.\n\n> C. Alternatively, would you like all parameters to come in on stdin,\n> credential helper style?\n\nThat's even worse. ;)\n\n> D. Other ideas?\n\nI really am happy just saying \"look, my script is basically openssh, so\njust assume that\". If it breaks, it breaks, and I'll fix it.\n\nThe one thing I was left puzzled by is why \"-G\" didn't work in the first\nplace, since the script really does just pass through its options.\nPoking around, I think the problem actually _is_ the old ssh version\nthing. I have a new one on my workstation, but our CI boxes are jessie,\nand have openssh 1:6.7p1-5+deb8u4. And that's where I was digging into\nthe failure.\n\nSo sorry for misleading you earlier. The wrapper script looks like it's\na red herring to some degree (I guess its name not being \"ssh\"\ncontributed, but then the -G detection failed).\n\n> If you were using an old version of OpenSSH, this would be a reason to\n> revive the old patch, but I'm tempted to stall longer just to get more\n> use cases like this to come out of the woodwork.\n\nSo apparently I am using an old version. Just switching to use \"-V\"\nseems like it might be another solution, then.\n\n-Peff\n"},{"id":"350240","messageId":"20180615072620.GA841@sigill.intra.peff.net","threadId":"47507","inReplyTo":"20180615051321.GA5840@sigill.intra.peff.net","subject":"Re: [ANNOUNCE] Git v2.16.0-rc0","fromName":"Jeff King","fromEmail":"peff@peff.net","sentAt":"2018-06-15T07:26:20Z","receivedAt":"2018-06-15T07:26:25Z","isPatch":false,"sender":{"key":"peff@peff.net","avatar":"https://avatars.githubusercontent.com/u/45925?v=4"},"body":"On Fri, Jun 15, 2018 at 01:13:21AM -0400, Jeff King wrote:\n\n> > The trouble is that GIT_SSH_VARIANT=simple is too... simple. You\n> > would like a variant that passes in [-p port] [-4] [-6] as well. We\n> > didn't implement that because we didn't have the attention of any\n> > wrapper writer who wanted it; in absence of a potential user, we\n> > decided to wait for a user to propose the interface they want. Now we\n> > can celebrate, since that day has come.\n> \n> I'm not sure I'm celebrating. It seems like work for not much benefit.\n> I'd just as soon use VARIANT=ssh and deal with any fallouts if my script\n> does not behave exactly like openssh in all regards.\n\nBy the way, if this sounds like I'm crapping all over your ideas, I\nreally don't mean to. What I'm trying to say is that I think your\noriginal patch may have hit a sweet spot between robustness and making\na simple-enough interface for users to configure.\n\n-Peff\n"}]}