> On 4 Sep 2024, at 17:34, David Rowley <dgrowleyml@gmail.com> wrote:
>
> On Wed, 4 Sept 2024 at 20:24, Daniel Gustafsson <daniel@yesql.se> wrote:
>> Not mandatory at all, but since you were prepping a typo backpatch anyways I
>> figured these could join to put a small dent in reducing risks for future
>> backports.
>
> I think this is pretty good logic. I think fixing comment typos in
> ancient code and backpatching to all supported versions isn't good use
> of time, but fixing a typo in "recent" code and backpatching to where
> that code was added seems useful. Newer code is more likely to need
> bug fixes in the future, so going to a bit more effort to make
> backpatching those bug fixes easier seems worth the effort.
Absolutely agree.
> I just don't know what "recent" should be defined as. I'd say if it's in a
> version we've not released yet, that's probably recent. By the time .1
> is out, there's less chance of bugs in new code. Anyway, I doubt hard
> guidelines are warranted here, but maybe some hints about best
> practices in https://wiki.postgresql.org/wiki/Committing_checklist ?
That sounds like a good idea. Off the cuff I would agree that unreleased
versions and .0 versions are strong candidates (but not mandatory) for trivial
backpatches like typos, beyond that the value is likely to be lower.
--
Daniel Gustafsson
