[Scummvm-devel] Memory consumption in GUI branch

[Vicent Marti]

Hey Max,

I've followed those exact same steps (twice) and the result is a  
complete mess:

It looks like SVNMerge is merging all the revisions on my branch to  
the trunk *one by one*, so for instance in a file I've developed from  
scratch (like xmlparser.cpp), it first merges the first revision into  
trunk, and then merges the second one on top of it creating a conflict  
between them. The end result is that all the files I've written are  
full of conflicts between their different revisions up to a point that  
they cannot be manually fixed.

This is very strange indeed... Maybe I'm missing something obvious?  
Maybe I trashed my branch sometime during the GSoC? That would indeed  
suck.

Cheers,
Tanoku

On Oct 5, 2008, at 10:00 PM, Max Horn wrote:

>
> Am 05.10.2008 um 20:44 schrieb Vicent Marti:
>
>> Fine with me. I'm willing to take all the blame when the trunk gets
>> trashed -- it's not like you actually know my *exact* location to  
>> bomb
>> me or anything.
>
> That's what you *think*... :)
>
>> I went ahead and tried a Svnmerge on my local copy... It turned out
>> rather bad because it showed conflicts on my own files (i.e. on files
>> that are only on the GUI branch and not in trunk)...
>
> Here's what I did when merging the RTL branch:
>
> 1) Made sure there were no local modifications in both my trunk and  
> my RTL branch checkout
> 2) Brought the RTL branch up to speed, i.e. performed a merge on it  
> agsinst trunk. That is:
>  cd gsoc2008-rtl
>  svnmerge merge -S /scummvm/trunk
>  [ ... manual cleanup of merge conflicts: test whether it all still  
> builds, etc. ...]
>  svn ci -F svnmerge-commit-message.txt
>
> 3) Then, merged everything back to trunk, i.e.
>  cd trunk
>  svnmerge merge --bidirectional -S /scummvm/branches/gsoc2008-gui
>  [ ... manual cleanup of merge conflicts, testing ...]
>
> 4) After I was dead sure the merge went right, I commited it with a  
> custom merge message.
>
> If you tell me what problems you got precisely at which step, maybe  
> we can work it out.
>
>
>> I either did something wrong or SVNMerge is even worse than what I  
>> had
>> expected. Any insight on this? I was hoping for a merge on Tuesday
>> afternoon, after I get everything sorted out. Who's going to stand up
>> and and do some watchful guidance (and some blame sharing, while  
>> we're
>> at it)?
>
> I can *try* to be there, but I can promise it at this time, as I  
> have to work hard on my thesis these days.
>
> Bye,
> Max