On Thursday, 28 July 2022 05:38:38 PDT Friedrich W. H. Kossebau wrote:
> Thanks for the link, interesting (Sidenote: Not sure I agree that there are
> SiCs which are acceptable, I know I got upset a few times by such breakages,
> breaking is breaking after all.)
They're all some level of bad. The question is what we could accept in some
circumstances that really require it, and those that we can't at all.
> Not sure I understand things correctly: Isn't the addition in our example
> rather category A, because it "can be worked around in user code without
> introducing [Qt] version checks"? In our example, by using A{}/A() or
> B{}/B() when invoking the method?
Strict reading of the QUIP would say it is Cat B because it adds overloads.
But they can be easily worked around in code without issues, so long as
obtaining the pointer to this function wasn't a requirement before, so more
like Cat A.
--
Thiago Macieira - thiago (AT) macieira.info - thiago (AT) kde.org
Software Architect - Intel DPG Cloud Engineering
