This is the mail archive of the
gdb-patches@sourceware.org
mailing list for the GDB project.
Re: [PATCH] Also install data-directory into the build directory as computed by relocate_gdb_directory
> From: Khoo Yit Phang <khooyp@cs.umd.edu>
> Date: Sat, 6 Oct 2012 15:36:13 -0400
> Cc: Khoo Yit Phang <khooyp@cs.umd.edu>,
> dje@google.com,
> brobecker@adacore.com,
> jan.kratochvil@redhat.com,
> gdb-patches@sourceware.org
>
> On Oct 6, 2012, at 3:24 PM, Eli Zaretskii wrote:
>
> >> From: Khoo Yit Phang <khooyp@cs.umd.edu>
> >> Date: Sat, 6 Oct 2012 15:02:00 -0400
> >> Cc: Khoo Yit Phang <khooyp@cs.umd.edu>, Joel Brobecker <brobecker@adacore.com>, Jan Kratochvil <jan.kratochvil@redhat.com>, GDB Patches <gdb-patches@sourceware.org>
> >>
> >> I think the bigger issue is that $BUILDDIR/gdb/data-directory overrides the standard data-directory. If we detect run-from-builddir based on the presence of other files/directories, and some other application happens to use the same files/directories, then the user is basically stuck with either a non-working gdb (sans -data-directory) or having to uninstall that other application.
> >
> > I see no reason to assume that we will not be able to reliably detect
> > a GDB build directory and to distinguish between that and other
> > projects. Surely, we can find at least one or 2 files that only exist
> > in GDB.
>
> But we have to make assumptions about other applications that are not under our control, that they do not use the same files, which I think is an assumption that we should avoid if possible (who knows what's out there).
How many other packages will have a file called doc/gdb.info or
doc/gdbint.info in their build tree? How many of them will have
something called i386-cygwin-tdep.o or mi/mi-cmds.o? I could find
many more files whose probability to be in a non-GDB tree is exactly
zero.
I think the problem you are afraid of doesn't exist.