[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[bug-gnulib] Re: xvasprintf module depends on xalloc-die

From: Jim Meyering
Subject: [bug-gnulib] Re: xvasprintf module depends on xalloc-die
Date: Sun, 27 Mar 2005 10:58:14 +0200

"Oskar Liljeblad" <address@hidden> wrote:
> On Friday, March 25, 2005 at 20:27, Jim Meyering wrote:
> [..]
>> > Shouldn't xalloc.h also be added to lib_SOURCES here?  Otherwise,
>> > xalloc.h might not be distributed when the maintainer does a "make
>> > dist".
>> Either that, or add m4/xvasprintf.m4 that does it
>> via AC_LIBSOURCES.  Either way is fine with me, but perhaps
>> someone (Oskar?) who actually uses the module should be involved.
> [..]
> Is there any gain in doing it with AC_LIBSOURCES? I must
> admit I'm not entirely acquainted with AC_LIB{OBJ,SOURCES,..}.

There is enough benefit in doing it that way that I've been converting
coreutils modules to use AC_LIBSOURCES and AC_LIBOBJ to specify their
dependencies on lib/*.[ch] files, and adding new .m4 files when necessary.
Recently, Paul merged those changes into gnulib.  Those changes make it so
that the .m4 file defines the dependency (which automake verifies), rather
than the relatively remote `Makefile.am:' section of the modules file.
The tighter coupling makes it so that that when I add a new module to
coreutils (which doesn't use gnulib-tool), I cannot forget to include
the corresponding .c and .h files -- it won't build without them, even
if those files would actually be used only on obscure systems.

> I thought these AC macros were to be used for compatibility
> matters?

They've been used in autoconf for ages to require the sources
corresponding to gnulib's error, obstack, and alloca modules.

reply via email to

[Prev in Thread] Current Thread [Next in Thread]