From mboxrd@z Thu Jan 1 00:00:00 1970 From: Zack Weinberg To: Ziemowit Laski Cc: "H . J . Lu" , gcc@gcc.gnu.org Subject: Re: Why is ENABLE_NLS always defined? Date: Tue, 16 Oct 2001 18:20:00 -0000 Message-id: <20011016182044.B17800@codesourcery.com> References: <20011016174902.B29772@lucon.org> <14AE11B0-C299-11D5-B89F-0030658361CA@apple.com> X-SW-Source: 2001-10/msg00940.html On Tue, Oct 16, 2001 at 05:51:21PM -0700, Ziemowit Laski wrote: > > On Tuesday, October 16, 2001, at 05:49 , H . J . Lu wrote: > > >The current gcc will enable NLS by default. It does check if > > exists. However, it still defines ENABLE_NLS even > >though it knows doesn't exist. That is on IRIX 6.5. > >I had to add --disable-nls. It doesn't happen with gcc on Oct. 5. > > Ah! Darwin has the same problem (I also have to --disable-nls now). --enable-nls is supposed to substitute our own gettext library if there isn't one in libc. However, our own gettext library uses a different header name from to avoid clashes, and the glue header (intl.h) doesn't know that. This should be fixed in the patch I sent last week: http://gcc.gnu.org/ml/gcc-patches/2001-10/msg00619.html zw