| 1 |
>>>>> On Thu, 18 Sep 2014, Tobias Klausmann wrote: |
| 2 |
|
| 3 |
> However, one aspect of how ebuilds are written these days will |
| 4 |
> cause a non-trivial amount of merge commits that are not actually |
| 5 |
> useful in that sense. |
| 6 |
|
| 7 |
> This is due to the way keywording and stabilization work on an |
| 8 |
> ebuild level. Since keywords are all in one line, any merge tool |
| 9 |
> will barf on two keywords being changed in disparate clones. I.e. |
| 10 |
> if I change ~alpha->alpha while someone else changes |
| 11 |
> ~amd64->amd64, we will have a merge conflict. |
| 12 |
|
| 13 |
> There several approaches to this problem: |
| 14 |
|
| 15 |
> 1) Make a merge commit as explained above. Aside from the |
| 16 |
> not really useful merge commit, this also means manual work for |
| 17 |
> the ATs, who could really do with less manual labor. |
| 18 |
|
| 19 |
> 2) roll back the local commit, fetch and re-do the semantically |
| 20 |
> equivalent commit. This might be automated, but someone has to |
| 21 |
> actually write that code. |
| 22 |
|
| 23 |
> 3) Do away with one-line KEYWORD lists. This has wide-ranging |
| 24 |
> repercussions, from grepping-and-parsing (which is already broken |
| 25 |
> in different ways[0]), and it makes the "preamble" of ebuilds |
| 26 |
> longer. |
| 27 |
|
| 28 |
If this should really turn out to be a problem, then we could also: |
| 29 |
|
| 30 |
4) Replace git's default merge driver by our own one that is better |
| 31 |
suited for ebuilds. This can be done per repository via .git/config |
| 32 |
and .gitattributes. |
| 33 |
|
| 34 |
Ulrich |
Archives