Luke Diamand <luke@xxxxxxxxxxx> writes:
> On Tue, 21 Jan 2020 at 23:05, Junio C Hamano <gitster@xxxxxxxxx> wrote:
>>
>> [jc] asking for help from those who made non-trivial changes to "git
>> p4" in the past 18 months or so for reviewing.
>
> This looks fine to me. I've not actually tested it.
>
> Ack.
Thanks, but it wasn't very helpful to see an Ack (i.e. "an expert
says this is good") without seeing any of my "why is this good?"
answered by either the original author or the expert X-<.
>> "Ben Keene via GitGitGadget" <gitgitgadget@xxxxxxxxx> writes:
>>
>> > From: Ben Keene <seraphire@xxxxxxxxx>
>> > Subject: Re: [PATCH] git-p4: Add hook p4-pre-pedit-changelist
>>
>> "git shortlog --no-merges" would show that the convention is to
>> downcase "Add".
>>
>> With two consecutive non-words (i.e. 'pre' and "pedit'), it really
>> feels an unpronounceable mouthful to a non-perforce person like me.
>>
>> On the core Git side, "git commit", which is the primary command
>> that is used to create a new commit, has two hooks that helps to
>> enforce consistency to the commit log messages:
>>
>> - The "prepare-commit-msg" hook prepares the message to be further
>> edited by the end-user in the editor
>>
>> - The "commit-msg" hook takes what the end-user edited in the
>> editor, and can audit and/or tweaks it.
>>
>> Having a matching pair of hooks and making sure the new hooks have
>> similar names to these existing ones may help experienced Git users
>> adopt the new hooks "git p4" learns here.
>>
>> What makes "p4-pre-pedit-changelist" a good name for this hook? "In
>> pure Perforce without Git, there is 'pre-pedit-changelist' hook that
>> Perforce users are already familiar with" would be a good answer but
>> not being P4 user myself, I do not know if that is true.
>>
>> Also, "git commit" has a mechanism (i.e. "--no-verify") to suppress
>> the "auditing" hook, and it serves as an escape hatch. The new hook
>> "git p4" learns may want to have a similar mechanism, to keep its
>> users productive even when they have broken/stale/bogus hook rejects
>> their legitimate log message, by allowing them to bypass the
>> offending hook(s).
>>
>>
>> > Add an additional hook to the git-p4 command to allow a hook to modify
>> > the text of the changelist prior to displaying the p4editor command.
>> >
>> > This hook will be called prior to checking for the flag
>> > "--prepare-p4-only".
>> >
>> > The hook is optional, if it does not exist, it will be skipped.
>> >
>> > The hook takes a single parameter, the filename of the temporary file
>> > that contains the P4 submit text.
>> >
>> > The hook should return a zero exit code on success or a non-zero exit
>> > code on failure. If the hook returns a non-zero exit code, git-p4
>> > will revert the P4 edits by calling p4_revert(f) on each file that was
>> > flagged as edited and then it will return False so the calling method
>> > may continue as it does in existing failure cases.
>>
>> The githooks(5) page should talk about some of these, I would think.
>>
>> > git-p4.py | 11 +++++++++++
>> > 1 file changed, 11 insertions(+)
>> >
>> > diff --git a/git-p4.py b/git-p4.py
>> > index 40d9e7c594..1f8c7383df 100755
>> > --- a/git-p4.py
>> > +++ b/git-p4.py
>> > @@ -2026,6 +2026,17 @@ def applyCommit(self, id):
>> > tmpFile.write(submitTemplate)
>> > tmpFile.close()
>> >
>> > + # Run the pre-edit hook to allow programmatic update to the changelist
>> > + hooks_path = gitConfig("core.hooksPath")
>> > + if len(hooks_path) <= 0:
>> > + hooks_path = os.path.join(os.environ.get("GIT_DIR", ".git"), "hooks")
>> > +
>> > + hook_file = os.path.join(hooks_path, "p4-pre-edit-changelist")
>> > + if os.path.isfile(hook_file) and os.access(hook_file, os.X_OK) and subprocess.call([hook_file, fileName]) != 0:
>> > + for f in editedFiles:
>> > + p4_revert(f)
>> > + return False
>> > +
>> > if self.prepare_p4_only:
>> > #
>> > # Leave the p4 tree prepared, and the submit template around
>> >
>> > base-commit: 232378479ee6c66206d47a9be175e3a39682aea6
