[Development] Trouble getting reviews in months and years

[Ilya Fedin]

On Fri, 11 Oct 2024 15:46:20 +0000
Volker Hilsheimer <volker.hilsheimer at qt.io> wrote:

> Hi Ilya,
> 
> 
> I’m sorry that your experience with contributing to Qt is less than
> stellar. Working on the platform integration of Qt, and perhaps in
> particular on the Xcb integration, is certainly challenging, for at
> least two reasons: the amount of variations of Linux systems; and the
> fact that much of that code is largely untested, and perhaps even
> untestable, through automatic testing. So regressions easily go
> unnoticed.
> 
> Given that, reviewers and platform maintainers are understandably
> very cautious with accepting patches to bugs that they can’t
> reproduce on the systems that they might have access to. There’s
> nothing wrong with that, even if it can make it an up-hill battle to
> get changes in that look obvious to you and are improving your
> particular experience with the framework.  

I would disagree that there's nothing wrong with that given that
changes like
https://github.com/qt/qtbase/commit/b71be292780b858f2c55ce92601452e2ea946de2
easily go in a patch update breaking input for a lot of people while a
change that can only affect applications explictly calling the method
it changes and adds code that present also in the same method of
Windows and Wayland QPAs that fixed the same issue on those platforms
gets a big resistance (speaking about 471045).

It's especially weird to have a bug report and not accepting patches
for it. If you're denying patches because you think there's no such
issue then the bug report should be closed, no? But it continues to be
open.

> 
> Claiming that people are “gaslighting you” is a strong accusation,
> implying malicious intent. From what I have seen, people are
> rejecting your patches because they can’t reproduce the bug, and are
> wary of the change breaking things on systems on which there are no
> problems today. Nobody is claiming that it’s your fault that Qt
> doesn’t behave the way you expect it to, and nobody is trying to
> manipulate you into thinking that you are imagining things.
> 
> So, I’d appreciate if you’d choose your claims a bit more carefully.  

I've used that termin not for the can't reproduce situation but for
situation where reviewers were claiming the same thing in a circle
without any argumentation. E.g. one reviewer said "KDE prefer
fontconfig.", I attempted to explain that it's not the reality and it's
enough to launch a KDE system and launch `xrdb -query` to see that X
resources font settings are set. Then he again says "Still disagree on
the basic premise.". What should I do with that? There are no
counter-arguments to break, just the reviewer insists on that even
though it's not the reality. How else should I call that?

(the reality is that those settings are equal to KDE and always
preferring X resources provides a way better end user experience as
applications start to behave consistently across environments
on X11 and even start to follow the settings in sandbox)