| 1 |
On Tue, Jul 25, 2017 at 4:29 PM, Mike Gilbert <floppym@g.o> wrote: |
| 2 |
> On Tue, Jul 25, 2017 at 12:12 PM, Michael Orlitzky <mjo@g.o> wrote: |
| 3 |
>> On 07/25/2017 09:23 AM, Michał Górny wrote: |
| 4 |
>>> |
| 5 |
>>> How is that relevant? Revision bumps are merely a tool to encourage |
| 6 |
>>> 'automatic' rebuilds of packages during @world upgrade. I can't think of |
| 7 |
>>> a single use case where somebody would actually think it sane to |
| 8 |
>>> checkout one commit after another, and run @world upgrade in the middle |
| 9 |
>>> of it. |
| 10 |
>>> |
| 11 |
>> |
| 12 |
>> Revisions are to indicate that one incarnation of a package differs from |
| 13 |
>> another in a way that the user or package manager might care about. And |
| 14 |
>> on principal, it's no business of yours what people want to do with |
| 15 |
>> their tree. If someone wants to check out successive commits and emerge |
| 16 |
>> @world, he's within his rights to do so. |
| 17 |
> |
| 18 |
> I don't feel I should be obligated by policy to support this use case. |
| 19 |
> One revbump per push seems sufficiently safe for 99.9% of users. |
| 20 |
> |
| 21 |
> If you want to do more revbumps, you are free to do so. |
| 22 |
> |
| 23 |
|
| 24 |
What is the point of separating changes by commits if we don't |
| 25 |
generally try to keep each commit working? |
| 26 |
|
| 27 |
Sure, there are some cases where it is just going to be too painful to |
| 28 |
ensure that, and so it doesn't have to be an absolute rule. |
| 29 |
|
| 30 |
However, if somebody is checking out a tree at some point in the past |
| 31 |
they shouldn't have to try to figure out where the last push boundary |
| 32 |
was to ensure that it is sane. Use cases for that include updating |
| 33 |
older systems progressively, or bisecting a problem. |
| 34 |
|
| 35 |
-- |
| 36 |
Rich |
Archives