Ignoring changed files

Steve Borho steve at ageia.com
Tue Dec 5 16:02:22 UTC 2006


On Monday 04 December 2006 23:28, Theodore Tso wrote:
> On Mon, Dec 04, 2006 at 05:31:54PM -0600, Steve Borho wrote:
> > It's such a common problem that I wonder if it should be enforced
> > internally.  Mercurial will prevent you from commiting any local
> > changes on top of a patch changeset.  Why not prevent the user from
> > pulling changes on top of a patched changeset?
>
> Pulling is perfectly safe.  If there is a mq-patch changeset applied,
> then an "hg pull" will result in the new changesets creating a new
> head revision.  It's only if you do an "hg merge" followed by an "hg
> commit" that you will run into problems, and the "hg commit" will
> result in an error.  Hg probably shouldn't allow you to do the "hg
> merge", though.
>
> In contrast, an "hg push" to a remote repository is far more
> dangerous, since the mq changeset is potentially emphemeral and could
> change after the next "hg qrefresh", the result is a mess --- and it
> gets even worse if other people pull down the change from the remote
> repository and then commit on top of it.

Yesterday I went ahead and added the hook work-around to the Wiki.  
What's the consensus on leaving it there?  If people think it's unhelpful I'll 
back it off.

http://www.selenic.com/mercurial/wiki/index.cgi/MqExtension

-- 
Steve Borho (steve at ageia.com)



More information about the Mercurial mailing list