Date: Fri, 6 Sep 2013 11:43:04 +0100 From: David Chisnall <theraven@FreeBSD.org> To: Konstantin Belousov <kostikbel@gmail.com> Cc: svn-src-head@FreeBSD.org, svn-src-all@FreeBSD.org, src-committers@FreeBSD.org Subject: Re: svn commit: r255297 - in head: include lib/libc/iconv lib/libiconv_modules Message-ID: <81E1EB28-814E-491B-B3C9-90B17CDF189D@FreeBSD.org> In-Reply-To: <20130906103636.GF41229@kib.kiev.ua> References: <201309060946.r869kj8T064459@svn.freebsd.org> <20130906103636.GF41229@kib.kiev.ua>
next in thread | previous in thread | raw e-mail | index | archive | help
On 6 Sep 2013, at 11:36, Konstantin Belousov <kostikbel@gmail.com> = wrote: > This is extremely rude. Such things are traditionally (and properly) > done with Makefile.inc in our tree. Look at the very beginning of > mk/bsd.init.mk. I was not aware of Makefile.inc, thank you. =20 > But the whole commit is hack. If you want to use C99 _Bool, use it = directly > instead of obfuscating the code through the build system. My intent was to minimise diffs from the citrus code. Our iconv.h leaks = stdbool.h into c89 files that use it, which is currently breaking around = a thousand ports (as would have been detected if there had been an exp = run before it was enabled by default). Removing this from the header = fixes the ports, but breaks the modules that expect bool to be defined. = This was the minimal change that would allow it to continue to build. The more elegant fix would be to modify all of the modules to explicitly = include stdbool.h if they wanted to use it, but this would then impose = lots more work on whoever does the next import from upstream. David (Who is currently trying to get the ports tree into a useable state for = 10.0)=
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?81E1EB28-814E-491B-B3C9-90B17CDF189D>