Instead of dying on specific use flags the ebuild should ignore cflags if custom-cflags is enabled for the ebuild and die otherwise. It is getting a bit old developers telling us the users what we can and can not build our package with. Reproducible: Always Expected Results: No ebuild should die for CFLAGS when custom-cflags is avaliable via a use flag, The ebuild can and should warn about unsafe cflags and state that support is not avaliable when using such cflags instead.
If you want to do that, you can still change the ebuild yourself, nobody forces you to use the original ebuild. What is in portage, is what is supported from us and we are not going to support insane CFLAG combinations
(In reply to comment #1) > If you want to do that, you can still change the ebuild yourself, nobody forces > you to use the original ebuild. What is in portage, is what is supported from > us and we are not going to support insane CFLAG combinations > And you wonder why people are getting tired of gentoo. You have no ideal about what CFLAGS are safe and unsafe. That is up to the user to decided. The use flag is there for this specific reason, refer to gaim ebuild and other such ebuilds that make use of the custom-cflags use flag. The all state they are not supported when the package is built with custom cflags which means you have no understanding of what the ideal or principle behind such a use flag is.
(In reply to comment #2) > And you wonder why people are getting tired of gentoo. You have no ideal about > what CFLAGS are safe and unsafe. That is up to the user to decided. Sigh; and you wonder why developers are tired of ricers? We have a very good idea about safe C[XX]FLAGS, unlike users. Search bugzilla for all the openoffice compile issues caused by ricing. > flag is there for this specific reason, refer to gaim ebuild and other such > ebuilds that make use of the custom-cflags use flag. The all state they are not > supported when the package is built with custom cflags which means you have no > understanding of what the ideal or principle behind such a use flag is. And people keep filing the bugs anyway, supported or not doesn't matter for them. The ebuild dies when you have -ffast-math set, because otherwise it *will* die (kaboom, bomb out) with a cryptic error after hours and hours of compile (see Bug 139412 and many duplicates). There's no way you could get it working w/ such flag. Using -ffast-math globally is immensenly stupid, read the fine man gcc. OO.org is *extremely* fragile wrt compiler flag. Finally please note that it's up to the maintainers to decide what they'll support or not, and custom-cflags is no way a standardized way to comfort ricers, OO.org being definitely the worst place to push for such flags and the two maintainers are already busy enough as it is, no need for them to deal with more pointless bugs.
