[svn.haxx.se] · SVN Dev · SVN Users · SVN Org · TSVN Dev · TSVN Users · Subclipse Dev · Subclipse Users · this month's index

Re: Subversion svn client doesn't show branch when checking in

From: Branko Čibej <brane_at_xbc.nu>
Date: 2006-03-08 19:02:59 CET

kfogel@collab.net wrote:
> Andrew Haley <aph@redhat.com> writes:
>
>> OK, I'm moving this to dev@subversion.tigris.org as it's a patch
>> submission.
>>
>> To recap: the svn client doesn't provide the full URL of a target file
>> when it opens the editor for "svn commit". Instead, it provides the
>> local file name. There's no way at that point to tell where in the
>> repository a file you'll be committing will go.
>>
>> Because of this, if I do "svn commit" in a private branch I get no
>> confirmation that files really will be committed to a private branch,
>> not (for example) the trunk.
>>
>> This is unnecessarily user-unfriendly, and it is a regression from
>> cvs, which does tell you which branches a file will be committed to.
>> It is true that you can use "svn info" before you use "svn commit",
>> but that's making extra work for the svn user.
>>
>> As far as I can see from the source code of svn/util.c, this is quite
>> deliberate: svn has all the information, but it is hidden from the
>> user.
>>
>> The only disadvantage I can see to applying this patch is that the
>> filenames in the editor will be longer.
>>
>
> Folks, I'm mildly inclined toward applying this patch, *if* we don't
> count the path style in the $EDITOR templates to be an API that can't
> be changed before 2.0. I don't think it's really an API -- programs
> that want the modified paths in a working copy in local style are most
> likely using 'svn status' anyway.
>
> Andrew has already pointed out the advantages. There is a slight
> disadvantage in that that one must transform URLs to local paths if
> one wants to use the local paths in the log message, but that's a
> largely automatable transformation,
Only if your whole working copy is consistently switched to a branch. If
only part of it is switched, or even only one file. then there's not
necessarily any correlation between the WC path and the repos path.

> and IMHO is outweighed by the
> advantage of having on-the-spot confirmation that one is indeed
> committing from the branch one thought one was. (Even we make the
> wrong-branch mistake sometimes, for example see r16946.)
>
> I haven't tested the patch yet, haven't made sure the item->url is
> present even for added-but-not-yet-committed files, etc (Andrew, have
> you tested that?). I'm just indicating tentative approval of the
> general idea, and soliciting comments from others. Thoughts?
>
If we add the target URL info to the template (and I'm not opposed to
that in general), then I'd suggest we show both the local path and the
URL. That means two lines per file, of course.

-- Brane

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Wed Mar 8 19:15:43 2006

This is an archived mail posted to the Subversion Dev mailing list.

This site is subject to the Apache Privacy Policy and the Apache Public Forum Archive Policy.