List of all commands that cause git gc --auto

问题:

Is there a definitive list of commands anywhere that cause git gc --auto to run? The git-gc(1) man page simply states:

--auto

With this option, git gc checks whether any housekeeping is required; if not, it exits without performing any work. Some git commands run git gc --auto after performing operations that could create many loose objects.

(emphasis added)

I'm in the process of organising a large migration from SVN to Git. The overwhelming majority of users will be on Windows PCs, and a not-insignificant portion of them are non-technical. They will be using TortoiseGit (as it closely matches TortoiseSVN, which they are already familiar with) - I've noticed that TortoiseGit does not include any functionality to run git gc manually at all.

The non-technical staff cannot be expected to have to launch a "git bash" command line to run git gc --auto from the appropriate working dir; and as we are using the "portable" distribution of MsysGit they will not have the "Git GUI Here.." windows shell context menu shortcut.

Is it reasonable to expect that over time Git will mostly self-maintain, or do I need to try and work out a non-technical user friendly method of invoking git gc --auto?

回答1:

builtin/merge.c:            const char *argv_gc_auto[] = { "gc", "--auto", NULL };
builtin/receive-pack.c:     "gc", "--auto", "--quiet", NULL,
git-am.sh:                  git gc --auto
git-rebase--interactive.sh: git gc --auto &&
git-svn.perl:               command_noisy('gc', '--auto');

From git grep -- --auto on git.git, those results looked interesting. The notable one is builtin/merge.c meaning that the ever so common git pull should trigger a git gc --auto.

Additionally, unless your 'non-technical' staff is doing rather 'advanced' stuff (at which point they wouldn't be 'non-technical' anymore), I don't see why they would ever need to run git gc manually instead of just letting git gc --auto handle everything.

回答2:

With Git 2.17 (Q2 2018), you will have to add git commit to the list of commands triggering a git gc --auto.
Actually, that should have been the case since the beginning of Git.

See commit 095c741 (28 Feb 2018) by Ævar Arnfjörð Bjarmason (avar).
^{(Merged by Junio C Hamano -- gitster -- in commit 9bb8eb0, 08 Mar 2018)}

commit: run git gc --auto just before the post-commit hook

Change the behavior of git-commit back to what it was back in d4bb43e ("Invoke "git gc --auto" from commit, merge, am and rebase.", 2007-09-05, Git v1.5.4-rc0) when it was git-commit.sh.

Shortly afterwards in f5bbc32 ("Port git commit to C.", 2007-11-08, Git v1.5.4-rc0) when it was ported to C, the "git gc --auto" invocation went away.

Since that unintended regression, git gc --auto only ran for git-am, git-merge, git-fetch, and git-receive-pack.
It was possible to write a script that would "git commit" a lot of data locally, and gc would never run.

One such repository that was locally committing generated zone file changes had grown to a size of ~60GB before a daily cronjob was added to "git gc", bringing it down to less than 1GB. This will make such cases work without intervention.

I think fixing such pathological cases where the repository will grow forever is a worthwhile trade-off for spending a couple of milliseconds calling "git gc --auto" (in the common cases where it doesn't do anything).