This is the mail archive of the
gdb-patches@sources.redhat.com
mailing list for the GDB project.
Re: [PATCH] Fix in-srcdir build of gcc
- From: Paolo Bonzini <paolo dot bonzini at lu dot unisi dot ch>
- To: DJ Delorie <dj at redhat dot com>, binutils at sources dot redhat dot com, gdb-patches at sources dot redhat dot com, GCC Patches <gcc-patches at gcc dot gnu dot org>
- Date: Sat, 26 Feb 2005 09:24:54 +0100
- Subject: Re: [PATCH] Fix in-srcdir build of gcc
- References: <421F4992.3030000@lu.unisi.ch> <200502251943.j1PJhnxo031954@greed.delorie.com>
As I am a proponent of "principle of least surprise"
I am too. Unfortunately this patch on one hand makes ./configure work
for gcc, on the other hand it makes it more "surprising" for
binutils/gdb. I am especially concerned about gdb, because of the two
it is the one that is most likely to be built by casual users doing
"./configure && make".
In-srcdir toplevel bootstrap definitely won't work without a patch such
as this, because it wants to symlink or rename directories (e.g.
stage2-gcc is linked to gcc, and stage1-gcc is linked to prev-gcc). But
maybe for the non-bootstrap case (and for gcc), it is enough on systems
that support symlinks to symlink the current directory to the host
directory. This would be better for the "principle of least surprise".
(Another remark on the changes to target libraries: newlib and winsup
are not affected because they do not special host tools, like fastjar
for libjava, and neither do they need to access tm.h in the gcc
directory, like libobjc).
Paolo