Re: git-p4: cloning with a change number does not import all files

From: Patrick Rouleau <prouleau72@gmail.com>
Date: Dec 2, 2017, 19:58 UTC
Message-ID: <CAL1_K8DGVzcbWFPz3vbEntB08sEhrV-j-sLxAh0QhV=8KRGf-Q@mail.gmail.com>
In-Reply-To: <CAE5ih78GOFqe_Vb3k_-jPiS7zbX1EdvYmsUyWvKKy=xOH9WrcQ@mail.gmail.com>

On Sat, Dec 2, 2017 at 12:55 PM, Luke Diamand <luke@diamand.org> wrote:

Show 7 quoted lines

> I think I've sort of stumbled across something like the problem you've
> described in the past. Perhaps the files you need have been deleted
> and then re-integrated or some such.
>
> Would you be able to take a look at some files with this problem and
> see if you can spot what's happened to it ("p4 changes" and perhaps
> "p4 changes -i").

Sorry, but these commands only show the date and an extract of the commit message.

Show 5 quoted lines

>
> One thing that can certainly happen is that Perforce gets *very*
> confused if you start getting too clever with symlinked directories,
> and git-p4 can only do so much in the face of this. But it may be
> something else.

Overall, the depot history is very "messy". It contains a lot of projects. The project on which I work has reach v5 and it has been branched from v4. From p4v's "show graph", I can see there was a 'main' branch at some point, but it doesn't exist anymore.

One of the files missing from my clone only has 2 revisions and it was created in the v5 branch. It was created at 608436 and modified at 608816. I cloned from 610443, mainly because we added a feature branch and I want to access it from git too (610443 is one month before the branch creation).

I will play a little bit with p4 to see if I can locate where the problem comes from and maybe hack git-p4 to make it more verbose.

Thanks, P. Rouleau