[QGIS-Developer] commits revert policy

[Sandro Santilli]

I'd like to call core developers and PSC attention on commits
revert policy. 3 commits of mine were reverted with this PR:

  https://github.com/qgis/QGIS/pull/57882

The PR was opened in my sleeptime and approved within 2 hours,
so I didn't have a chance to review the PR, which I think does
not help building a collaborative team around QGIS development.

I take responsibility on the contribution I provide and I think
that should be a requirement for being "a committer" (meaning of
which is yet to be defined since the switch to git).

Before switching to git being a "committer" meant taking a
committment to take care of the code, as opposed to send a patch
and disappear forever. I'd like to hold to my responsibilities
and thus partecipate in a fix of a problem I introduced.

This kind of responsibility is the one I see for example in
Benoit when he reviewed a PR that was reverting his commits here:

  https://github.com/qgis/QGIS/pull/57802

My reverting PR had the approval of the commit author, although
that one did not count as per the currently enforced rules (which
are inappropriate, IMHO, in this specific case).

Instead in the reverting of my commits I could not have a say,
and I don't think it's acceptable, especially when the revert is
said to be required to fix a segfault while no detail was given
about that. CI was green so did not trigger the segfault and
all the reverted commits did was avoiding hard-coded paths and
consistently honouring TMPDIR environemnt variable.

To me, if honouring TMPDIR triggers a segfault we clearly have a
bug in the testsuite, and I think an effort should be made (by all
developers) to help with fixing the testcase, not hiding the bug
under the carpet by disabling the test.

With this mail I propose the following policy:

  1. If a bug is found, an issue is filed with all the details
     about how to reproduce it

  2. If an envisioned solution involves reverting a commit,
     the commit author is given enough time to review it (1 week?)

  3. Core committers should be ready to take responsibility on
     the changes the propose, failing which should be considered
     for removing the "commit" privileges

In addition to the above, what it means to be a "committer" today
needs be defined on the governance page as it isn't clear by the
sole name, because currently I think what it means to be a
"committer" is simply to have the privilege of approving other
people PRs (but not your own). This should be also revistited, IMHO,
but it's better handled in a separate thread.


--strk; 

  Libre GIS consultant/developer
  https://strk.kbt.io/services.html
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 659 bytes
Desc: not available
URL: <http://lists.osgeo.org/pipermail/qgis-developer/attachments/20240626/9f378d08/attachment.sig>