[devel at bb.net] Improved P4 (Perforce) support

[Dustin J. Mitchell]

It all sounds fairly perforce-specific, and I'm not a perforce user.  But
as long as it doesn't interfere with existing perforce behaviors, I don't
see any reason not to merge it.

Dustin

On Fri, Feb 5, 2016 at 8:25 AM, Vedran Krivokuca <vkrivokuca at gmail.com>
wrote:

> Hi all,
>
>
> I have made a slight modification of one of the source steps in
> Buildbot, concerning support for Perforce streams. My idea is to now
> make pull requests for both master and eight branch (have separate
> patches prepared for that), but I would like your input if you want it
> contributed some other way.
>
> Streams are "light-weight" branching mechanism which resemble
> Git-style branching more than the "old" "SVN-style" depots and
> branches which were the default approach with Perforce few years ago.
>
> On Perforce Streams in general:
>
> http://www.perforce.com/perforce/doc.current/manuals/p4v/streams_overview.html
>
> So, why is this important?
>
> Streams are more of a "virtual" storage mechanism in Perforce, they
> can reference parent streams only partially, can re-map parts of the
> parent streams and so on... When replicating a Stream just like a
> regular Depot (which current implementaiton of P4 source step always
> doesi n Buildbot), these changes can reflect partially.
>
> As an illustration, please observe attached graph of Stream depencies
> on a live development project:
>
>  - Box A represents mainline stream
>  - Boxes B, C, D (and all of their same-level counterparts) represent
> various feature/development streams
>  - Box E represents a virtual stream, which remaps parts of the folder
> hierarchy from the stream E to reflect needs of the stream in the box
> F
>  - Finally, box F represents completely new development project, which
> relies on the parts of the source code from the stream in the box D
>
> With the current implementation of the Perforce support in Buildbot,
> box A always works fine (can be directly 1:1 replicated as if it's
> being an old-style Perforce depot).
>
> Boxes B-D can backfire, depending on the complexity of their setup (i.
> e. if they have remapped or ignored paths).
>
> Boxes E and F don't work at all properly, because features of such
> streams are not recognized correctly if stream is pulled as depot.
>
>
> Do I need to explain the origin of the need for this improvement
> further, or should I make pull requests?
>
>
> Best regards,
> Vedran.
>
>
>
> --
>   Pozdrav/Greetings,
>   Vedran Krivokuća
>   Disclaimer: This message may contain information.
>
> _______________________________________________
> devel mailing list
> devel at buildbot.net
> https://lists.buildbot.net/mailman/listinfo/devel
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.buildbot.net/pipermail/devel/attachments/20160207/d6baeabb/attachment.html>