On 5/10/12 5:13 PM, "ext Olivier Goffart" <[email protected]> wrote:
>On Thursday 10 May 2012 07:10:51 [email protected] wrote: >> On 5/10/12 4:20 AM, "ext Thiago Macieira" <[email protected]> >> >> wrote: >> >On quarta-feira, 9 de maio de 2012 21.39.53, [email protected] >>wrote: >> >> How about this then? >> >> >> >> https://codereview.qt-project.org/#change,25788 >> >> >> >> https://codereview.qt-project.org/#change,25787 >> > >> >Yes, something like that. I'm not sure we should do it on 5.0 though. >> >> At least as an option I'd like to have it for 5.0. But I can't see how >>it >> can break things given that QtCore has exceptions enabled. In addition, >>we >> disable them only when compiling Qt modules, not when compiling app >>code. >> So I would rather try it for the beta and see whether we get any >>feedback >> about issues with it. >> >> My only concern is whether enabling/disabling exceptions changes the ABI >> on any platform/compiler. > >At least with GCC, this should be safe (provided applications don't let >exceptions go trough the Qt layer) Yes, it's safe with gcc/ELF, I was more wondering about MSVC. >There is also QtConcrrent which is not part of QtCore anymore and has >some >exception handling possibilities. At least currently it doesn't use any try/catch statements. > >And I think -no-exception could stay to disable exception even in QtCore >if >someone still want to reduce the footprint I'm not sure it's worth it. We probably gain more by removing one more configuration option that we'd otherwise need to maintain and test. If we really want to remove that overhead the better option would probably be to only compile the files that need it with exceptions enabled. Cheers, Lars _______________________________________________ Development mailing list [email protected] http://lists.qt-project.org/mailman/listinfo/development
