| 1 |
On 11/06/2015 08:34 AM, Kent Fredric wrote: |
| 2 |
> On 7 November 2015 at 02:16, Michael Orlitzky <mjo@g.o> wrote: |
| 3 |
>> These days, if I'm careful to revbump when necessary AND limit my |
| 4 |
>> commits to one logical change, can I wind up going from (say) -r1 all |
| 5 |
>> the way to -r4 before pushing my changes. |
| 6 |
> |
| 7 |
> Personally I don't think that's necessary. The "-r bump on dep change" |
| 8 |
> argument is a defence against installer limitations and the |
| 9 |
> replication of changes to users. |
| 10 |
> |
| 11 |
> As your commits were never pushed to users to install from, you've not |
| 12 |
> made a change that can visibly impact a single person, at least, not |
| 13 |
> under "typical" usecases. |
| 14 |
> |
| 15 |
|
| 16 |
Sure, but the whole point of using logical commits is that somebody |
| 17 |
might need to revert or check out one commit in a series. If you need to |
| 18 |
undo an r5 -> r6 bump that modified RDEPEND, then you want the revision |
| 19 |
to go back to r5, too, so that users rebuild it. |
| 20 |
|
| 21 |
|
| 22 |
>> |
| 23 |
>> What magic is this? And how do I make it my default in gentoo.git? |
| 24 |
> |
| 25 |
> There's a lot of different parameters that take effect here: |
| 26 |
> |
| 27 |
> --find-copies-harder |
| 28 |
> -M |
| 29 |
> -C |
| 30 |
> -B |
| 31 |
> |
| 32 |
> All have different effects on the verbosity and/or performance in |
| 33 |
> different situations. |
| 34 |
|
| 35 |
Niceeeeee. Thanks to you and Peter, I figured out that I can get away |
| 36 |
with this in my .gitconfig: |
| 37 |
|
| 38 |
[diff] |
| 39 |
renames = true |
| 40 |
|
| 41 |
If that ever doesn't work, --find-copies(-harder) should help. |
Archives