[Development] Requesting Feature Freeze Exception for C++20 comparison

[Thiago Macieira]

On Monday, 5 June 2023 23:45:03 PDT Alex Blasche via Development wrote:
> The point of using date and time as example was to proof the concept. This
> is an API offer for other maintainers to adopt in their API offering. We
> have never made it a condition for features to be adopted by every single
> module.

My point wasn't that rolling out features to other submodules was a 
requirement. It's instead that I don't think the 4 classes from date/time plus 
qfloat16 that I asked are enough of a proof of concept. And I have serious 
concerns about binary compatibility.

> The patch series was up for review and there was a very active discussion
> including naming for weeks already. While I accept concerns I am very
> disappointed that such reviews come in 2 days before FF with immediate
> rejection for the upcoming release due to timing. This was not the first
> time. Since this diminish my confidence that this won't happen again I
> would like to see a measure in place that can avoid this.

Sorry, but I was busy (and still am). I have 2 or 3 hours per week to dedicate 
to Qt, including writing my own contributions, reacting to reviews by others 
to them (see the QProcess changes), and reviewing other people's 
contributions. Plus the security team actions, which preempt everything else. 
So I have to prioritise.

In particular to this change, I also thought it was important to find out 
whether we needed it at all, or whether we could get the C++20 bump first. So 
we began that discussion and concluded that we needed C++17 compatibility.

I don't complain that the my changes take 1 to 2 months to get merged (e.g.,  
the QProcess::setUnixProcessParameters() change was first submitted on March 16 
and merged on May 22). It's annoying that I have to send 1- and 2-week pings 
to get reviews, but I accept that everyone is busy.

I'd much rather get a review that improves a feature, even if it's on the last 
minute, than to accept a broken feature. The quality of the review does not 
depend on the time when the review came in. If it had come in sooner, maybe we 
could have got this in, but so what? Is this feature really that important to 
get in now?

-- 
Thiago Macieira - thiago.macieira (AT) intel.com
  Cloud Software Architect - Intel DCAI Cloud Engineering
-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/pkcs7-signature
Size: 5152 bytes
Desc: not available
URL: <http://lists.qt-project.org/pipermail/development/attachments/20230606/fedc8505/attachment.bin>