[gmx-developers] Git master broken

[Christoph Junghans]

Hi,

the merge was not 100% trivial, there were two simple conflicts
due to a define renaming in include/vec.h and a argument change in
split of include/string2.h, which only happened in the master branch.
That is why I committed it with the standard commit message.

I just did the merge, because the master branch missed a header of some 
kernel, which was fixed in 4.5 branch. After the merge I did a build 
check and it worked, so this was not a build break.


Concerning the current merge policy:
I think the problem by now is that the main work happens on the 4.5 
branch and this has to be merged into the master branch immediately. 
Otherwise the first one, who wants to merge something in the master 
branch, in the above case me, has to decide about all the 4.5 changes of 
all the other developers. At the current point of development this is 
not a big deal, because master and 4.5 are not very different.

However, what still puzzles me is the origin of the conflicts in the 
case of David's merge. For human eyes most of them look trivial, but git 
doesn't get it.

Cheers,

Christoph

Am 25.08.2010 16:53, schrieb David van der Spoel:
> On 2010-08-25 16.45, Mark Abraham wrote:
>>
>>
>> ----- Original Message -----
>> From: David van der Spoel <spoel at xray.bmc.uu.se>
>> Date: Wednesday, August 25, 2010 23:58
>> Subject: [gmx-developers] Git master broken
>> To: Discussion list for GROMACS development <gmx-developers at gromacs.org>
>>
>> > Hi,
>> >
>> > it seems that git master is broken. Try the following:
>> >
>> > mkdir koko
>> > cd koko/
>> > git clone git at git.gromacs.org:gromacs
>> > git checkout -t origin/release-4-5-patches
>> > git merge release-4-5-patches
>>
>> That just worked fine for me.
>>
>> I get your symptoms if I do "git checkout master" before the merge,
>> whether or not I checkout release-4-5-patches.
>>
>> Christoph Jungan's merge commit on master dated Monday (5e3473a12) is
>> not a simple merge commit (see git log -p). I'm not sure how he managed
>> to do what he did, but I guess he started a merge from some
>> not-quite-right starting point, had conflicts, didn't realise he'd not
>> commited the merge result, did his edit, then committed.
>>
>> To fix, someone will have to manually resolve the conflicts.
>>
>
>
> OK I will do this now.
>
>> Mark
>>
>> > less include/resall.h
>> > <<<<<<< HEAD
>> > t_restp *search_rtp(const char *key,int nrtp,t_restp rtp[]);
>> > /* Search for an entry in the rtp database */
>> > =======
>> > char *search_rtp(const char *key,int nrtp,t_restp rtp[]);
>> > /* Search for an entry in the rtp database, returns the rtp
>> > residue name.
>> > * A mismatch of one character is allowed, if there is only
>> > one nearly
>> > * matching entry in the database, a warning will be generated.
>> > */
>> >
>> > t_restp *get_restp(const char *rtpname,int nrtp,t_restp rtp[]);
>> > /* Return the entry in the rtp database with rtp name rtpname.
>> > * Generates a fatal error when rtpname is not found.
>> > */
>> > >>>>>>> release-4-5-patches
>> >
>> > And a few more conflicts.
>> >
>> > Any clue how to fix this?
>> >
>> >
>> > --
>> > David van der Spoel, Ph.D., Professor of Biology
>> > Dept. of Cell & Molec. Biol., Uppsala University.
>> > Box 596, 75124 Uppsala, Sweden. Phone: +46184714205.
>> > spoel at xray.bmc.uu.se http://folding.bmc.uu.se
>> > --
>> > gmx-developers mailing list
>> > gmx-developers at gromacs.org
>> > http://lists.gromacs.org/mailman/listinfo/gmx-developers
>> > Please don't post (un)subscribe requests to the list. Use the
>> > www interface or send it to gmx-developers-request at gromacs.org.
>>
>
>

-- 
Dipl.-Phys. Christoph Junghans
Max Planck Institute for Polymer Research
Theory Group
POBox 3148
D 55021 Mainz, Germany

Phone: +49 6131 379 335
Web: http://www.mpip-mainz.mpg.de/~junghans