This is the mail archive of the
libstdc++@sources.redhat.com
mailing list for the libstdc++ project.
Re: Again: Bootstrap failure for objdir containing colon
- To: Gerald Pfeifer <pfeifer at dbai dot tuwien dot ac dot at>
- Subject: Re: Again: Bootstrap failure for objdir containing colon
- From: Alexandre Oliva <aoliva at redhat dot com>
- Date: 16 Dec 2000 13:57:15 -0200
- Cc: <gcc at gcc dot gnu dot org>, <libstdc++ at gcc dot gnu dot org>, Geoffrey Keating <geoffk at redhat dot com>
- Organization: GCC Team, Red Hat
- References: <Pine.BSF.4.30.0012161602470.44156-100000@deneb.dbai.tuwien.ac.at>
On Dec 16, 2000, Gerald Pfeifer <pfeifer@dbai.tuwien.ac.at> wrote:
> Well, I think there is a big difference here: We can easily control
> whether we use colons in the source directories or not, but we cannot
> easily control the names object directories.
???
I don't see any difference here. What's the difference between your
storing your build tree in /tmp/`date`/gcc and my storing my source
tree in ~/src/`date`/gcc-cvs? Any Makefile target that refers to
$(srcdir) will break, then.
> This problem has been fixed two times during the last month and now
> reappeared for a third time, and the failure really was a bit hard for
> me to understand, so just imagine some regular user. :-O
It was never really fixed. All attempted fixes broke something else.
As I said, it's tricky to fix correctly.
> What confuses me a bit is why this is causing problems for libstdc++-v3
> while it has been working for years (and the rest of GCC). Is there some
> special requirement libstdc++-v3 has?
Because libstdc++-v3 is the first package (AFAIK) that exports its
build parameters (header and library search paths), as opposed to
having them hard-coded in the top-level build machinery.
--
Alexandre Oliva Enjoy Guarana', see http://www.ic.unicamp.br/~oliva/
Red Hat GCC Developer aoliva@{cygnus.com, redhat.com}
CS PhD student at IC-Unicamp oliva@{lsd.ic.unicamp.br, gnu.org}
Free Software Evangelist *Please* write to mailing lists, not to me