[PATCH 4 of 5] merge.mergestate: store whether the merge is a branch merge

[Martin von Zweigbergk]

On Tue, Nov 3, 2015 at 5:28 PM Siddharth Agarwal <sid at less-broken.com>
wrote:

> On 11/3/15 17:19, Martin von Zweigbergk wrote:
> > I instead used a 'T' field to indicate the type of conflict, which
> > could be 'm', 'cd' or 'dc'. It's been too long for me to understand
> > what the pros and cons of that would be versus your approach of
> > storing the branch merge flag. What do you think?
>
> My future patches will introduce something similar to that as well. The
> reason the branchmerge flag needs to be stored is related to dirstate
> modifications that need to be done based on whether this is a branch
> merge. For example, a local changed/remote deleted file needs to be
> marked added iff this isn't a branch merge.
>

Oh, I see. I see that I did "if repo.dirstate.p2() != nullid" in my patch.
Would that be better or worse?


>
> >
> > Here is my attempt, rebased part of the way to the current @:
> >
> http://42.netv6.net/martinvonz-wip/mercurial/log?rev=197a4ea5f82d%3A%3A89aa1f9405a5
> .
> > I don't even remember if there was some flaw in it or why I didn't
> > send it for review back then. It might be that the problems are now
> > fixed by the patches dealing with .hgsubstate that I saw a while back.
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://selenic.com/pipermail/mercurial-devel/attachments/20151104/0c747a62/attachment.html>