Merges lost

[Tim Armes <ta...@fr.imaje.com>]

We seem to be having an occassional (and unreproducable problem) whereby a
developer will update their local working copy which results in files being
merged automatically.  Once committed however, it becomes apparent that the
merged files are actually not merged, and that the version now in the
repository is that of the developer - i.e. the previous changes made before
his update are no longer in the head revision.

For the moment this is all the information that I have.  Is this a known
problem?

Tim
###########################################

This message has been scanned by F-Secure Anti-Virus for Microsoft Exchange.
For more information, connect to http://www.F-Secure.com/

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org

Re: Merges lost

[Michael W Thelen <th...@cs.utah.edu>]

* Tim Armes <ta...@fr.imaje.com> [2004-04-01 09:52]:
> We seem to be having an occassional (and unreproducable problem) whereby a
> developer will update their local working copy which results in files being
> merged automatically.  Once committed however, it becomes apparent that the
> merged files are actually not merged, and that the version now in the
> repository is that of the developer - i.e. the previous changes made before
> his update are no longer in the head revision.

I would strongly suspect that this is a user problem, perhaps the text editor
behavior mentioned by Lorenz Minder.  Do your developers run 'svn diff' before
committing to make sure they are committing exactly the changes they think
they are?  This is a good policy in general, and would help to make it clear
whether the problem is really a user problem or a Subversion problem.

-- Mike

-- 
Michael W. Thelen
Under capitalism, man exploits man.  Under communism, it's just the opposite.
                -- John Kenneth Galbraith

Re: Merges lost

[Lorenz Minder <lm...@gmx.net>]

Tim Armes wrote:
> We seem to be having an occassional (and unreproducable problem) whereby a
> developer will update their local working copy which results in files being
> merged automatically.  Once committed however, it becomes apparent that the
> merged files are actually not merged, and that the version now in the
> repository is that of the developer - i.e. the previous changes made before
> his update are no longer in the head revision.
> 
> For the moment this is all the information that I have.  Is this a known
> problem?

Hello!

I have sometimes seen this happen with people using texteditors which
don't warn when trying to save files which have changed on disk since 
loaded. Consider the following:

* User 1: checkout/update
* User 2: checkout/update
* User 1: edit, save, commit
* User 2: edit, then "svn update" in seperate terminal
	-> User 1's changes are locally merged
* User 2: continues editing w/o having reloaded the file in the 
editor 			(Hence, he still edits the old, unmerged version.)
* User 2: save, commit
	-> User 1's commit has gone.

Yours,
--Lorenz


---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org

Re: Merges lost

[Ben Collins-Sussman <su...@collab.net>]

On Thu, 2004-04-01 at 03:13, Tim Armes wrote:
> We seem to be having an occassional (and unreproducable problem) whereby a
> developer will update their local working copy which results in files being
> merged automatically.  Once committed however, it becomes apparent that the
> merged files are actually not merged, and that the version now in the
> repository is that of the developer - i.e. the previous changes made before
> his update are no longer in the head revision.
> 
> For the moment this is all the information that I have.  Is this a known
> problem?

Never heard of a such a problem, ever.  And it's hard for me to believe,
honestly, because a bug this fundamental should have been easily spotted
over the course of four years of development and testing.  

My suspicion is that the symptom of your problem (losing local
modifications) is coming from a complex mixture of user actions which
result in accidental data loss.  (PEBKAC, in other words.)  But hey, I
could be wrong.  It would be fantastic if you can find a way to
reproduce.



---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org