incubator-ooo-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Mathias Bauer <Mathias_Ba...@gmx.net>
Subject Re: A point on the OOo code -- Getting to Apache Extras
Date Wed, 31 Aug 2011 07:38:36 GMT
Hi Dennis,

Am 30.08.2011 05:26, schrieb Dennis E. Hamilton:

> Mathias,
> 
> In this thread you mentioned Apache Extras.  I am curious how to
> support that moving ahead.  So I must confess my ignorance first.
> 
> Have you considered what a roadmap would be to have the CWS that need
> preservation moved to Apache Extras?
> 
> I know we don't need everything in the CWS segment of
> http://hg.services.openoffice.org, and I think there is a list of the
> serious ones somewhere earlier on ooo-dev.
> 
> But since there are many, what is an appropriate way to get them onto
> Apache Extras?
> 
> I've seen you say that there needs to be a combined repo, where they
> are bookmarked.  I'm not sure what it means to have a combined repo
> based on the trunk repo and I'm unclear what happens when two CWS
> have changes to the same file?  Are collisions not possible or do we
> have to resolve them in having the combined repo?
> 
> I have more questions but this is too many already.
> 
> - Dennis
> 
> PS: I'm assuming the combination is done because we only get one Hg
> repo in a single Apache Extra project and we also want to reduce the
> size of the combined collection of CWS.
> 
> PPS: Maybe an easier question to ask is, what is the procedure to
> have the effect of cloning two CWS and merging them into one Hg repo
> with them bookmarked separately?  Is this straightforward or
> complex?

First of all, your assumption is correct: putting all cws into one
repository boils down the whole repository a lot. And it is not a very
complicated procedure.

My suggestion would be to create such a repository, export it into a
Mercurial bundle and import that into a Mercurial repo at the Apache
Extras site. Admittedly I don't know how this site is used, but as my
suggestion would be a quite easy and straightforward task, I hope that
it is possible to do it that way.

Collisions won't happen in that repository, it just contains the
changesets of all cws we have and tracks their history.

Conflicts could happen if we merged one of the heads representing the
cws with the "master" code line. But I wouldn't recommend to do that. We
should use this Mercurial repo only as a source for further work that
will happen elsewhere.

We really should convert the cws into patch queues, apply these patches
to new svn branches and fix the conflicts there. Michael Stahl has
created a script to convert cws into patch queues and he already has
tested it with some quite complicated cws.

Regards,
Mathias


Mime
View raw message