| 1 |
On 6 June 2016 at 05:34, rindeal <dev.rindeal@×××××.com> wrote: |
| 2 |
> efore merging to master they do a rebase+force push, |
| 3 |
> let the CI check it, and if it passes they're allowed to merge it. No |
| 4 |
> cherry-picking involved. Master is then always clean and happy. I'm |
| 5 |
> not sure how well it scales, but for the 100-150 commits/day the tree |
| 6 |
> sees currently, it should be doable. |
| 7 |
|
| 8 |
|
| 9 |
And then somebody else merges in the interval between you submitting |
| 10 |
it, the CI check starting, and the CI check approving it (keeping in |
| 11 |
mind the CI run is ~10 minutes long), but what merged breaks your |
| 12 |
commit and the CI didn't see it? |
| 13 |
|
| 14 |
Now what? |
| 15 |
|
| 16 |
Or does everyone have to form an orderly queue where they wait for the CI ? |
| 17 |
|
| 18 |
That' maxes out at 144 pushes / day. |
| 19 |
|
| 20 |
This is not the kernel where we have a clean release cycle we can |
| 21 |
target and sit on patch series for a week or more while we polish it, |
| 22 |
and we don't have an avalanche of "this is good, it lands in this |
| 23 |
release". |
| 24 |
|
| 25 |
This is pretty much one of the big limitations of "Shared push" |
| 26 |
patterns in that their workflow and "unified branching" workflows |
| 27 |
contradict each other in lots of ways. |
| 28 |
|
| 29 |
-- |
| 30 |
Kent |
| 31 |
|
| 32 |
KENTNL - https://metacpan.org/author/KENTNL |
Archives