This is the mail archive of the
egcs-patches@egcs.cygnus.com
mailing list for the EGCS project.
Re: libgcj versus libtool
- To: Alexandre Oliva <oliva@dcc.unicamp.br>
- Subject: Re: libgcj versus libtool
- From: Tom Tromey <tromey@cygnus.com>
- Date: Fri, 2 Jul 1999 18:17:09 -0700
- Cc: tromey@cygnus.com (Tom Tromey), egcs-patches@egcs.cygnus.com, java-patches@sourceware.cygnus.com, libtool@gnu.org
- References: <orwvwnrrmd.fsf@saci.lsd.dcc.unicamp.br><199906291701.KAA23765@ferrule.cygnus.com.><orso79oocm.fsf@saci.lsd.dcc.unicamp.br><m10zOxs-000NVRC@creche><ork8sj8jrt.fsf@cupuacu.lsd.dcc.unicamp.br>
>>>>> "Alexandre" == Alexandre Oliva <oliva@dcc.unicamp.br> writes:
>> Here is the rather ugly patch. You'd probably want to remove the
>> "CYGNUS LOCAL" markers.
Alexandre> Hmm, the problem is that it assumes gcc. -B may have other
Alexandre> meanings, and we don't always want to pass it down.
This patch is really only any good for the Cygnus (or egcs) tree. I
wouldn't recommend it for general use.
In the egcs tree we can probably assume that -B will only be used in
this particular way.
Alexandre> This is already implemented in libtool 1.3b, that is likely
Alexandre> to be released today. I understand it might be cumbersome
Alexandre> to add -Wc, before -B, but I don't see any other portable
Alexandre> way to go :-(
Doesn't this end up being the same as your previous patch? We have to
change CC in a particular way on a case-by-case basis in both
scenarios.
It is a bad situation. No approach seems very clean. Perhaps going
with your "-lt-" patch, and adding dummy dependencies for
compatibility, is the best way. It adds a bit more overhead to a file
that is already complex. To me this seems a bit better than
maintaining a local hack to libtool.
Tom