[prev in list] [next in list] [prev in thread] [next in thread] 

List:       pykde
Subject:    Re: Issue with PyQt6 on PyPI
From:       Phil Thompson <phil () riverbankcomputing ! com>
Date:       2024-04-26 11:20:57
Message-ID: d724b2827a8f92f785d5e9590be87e53 () riverbankcomputing ! com
[Download RAW message or body]

On 26/04/2024 12:11, Florian Bruhin wrote:
> Hey,
> 
>> They take advantage of the binary compatibility "promises" of Qt so
>> that you can use, for example, PyQt6 v6.x with Qt6 v6.x+1. However
>> those "promises" cannot be relied on these days
> 
> FWIW, while accidental ABI breaks definitely have happened in Qt 
> before,
> I don't think the blame is on Qt for this one.
> 
> QFont::tagToString() was marked preliminary in Qt 6.6:
> https://doc.qt.io/archives/qt-6.6/qfont.html#tagToString
> 
> While I'm struggling to find this written down anywhere, it seems
> reasonable to assume ABI breaks can happen in API marked as 
> preliminary.
> 
> I wonder if it'd make sense for PyQt to avoid exposing such API (or at
> least hide it between some build flag which is disabled by default)
> until Qt declares it stable.
> 
> Florian

I agree that was a bug - and had I imposed an upper limit to the version 
of the Qt6 dependency it wouldn't have mattered.

There are similar new experimental methods in existing classes in Qt 
v6.7 which I have not implemented.

Phil
[prev in list] [next in list] [prev in thread] [next in thread]