[Development] Disavowing the Lakos Rule for Q_ASSERT

[Turtle Creek Software]

Most of this discussion is way beyond my C++ skill level.  But I think the
root problem may be that Qt has (at least) two very different user classes.

For users working on mission-critical embedded systems, code rigor is
important.  Performance is important.  It's worth worrying about the
performance gains that noexcept might provide.

For those building a GUI consumer app, UX is paramount. The main focus for
exception design is to make things as easy as possible for the end user. A
nanosecond or even a millisecond difference is not even visible. The finer
points of C++ aren't an issue, since 99.99% of code is just widgets doing
simple things.

Maybe Qt would benefit from a strictness setting, especially for exception
handling?

I mentioned in previous posts that it's important for us to swallow
exceptions in the event loop so users can continue work after rare errors.
Extremely non-rigorous, but it's just someone's data entry screen, not a
medical device or power plant.  Worst case the user has to force-quit, no
worse than a throw out of the loop.

If an option to relax some rigor allowed that to happen reliably, it would
be a very good thing for us.

Best, Casey McDermott
TurtleSoft.com
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.qt-project.org/pipermail/development/attachments/20240828/d4f85016/attachment-0001.htm>