On 7/23/07, M.Canales.es <[EMAIL PROTECTED]> wrote: > El Lunes, 23 de Julio de 2007 02:37, Dan Nicholson escribió: > > > That's what I meant. > > tst-vfork3.out just contains: > > script 1 > script 1 > script 1 > script 1 > script 1 > script 2 > script 2 > script 2 > script 2 > script 2 > script 3 > script 3 > script 3 > script 3 > script 3 > echo failed with status 512 > > Do you need tst-vfork3.mtrace and/or test-vfork3.o.d?
That doesn't say too much. OK, looking at postix/test-vfork3.c, I think I see the issue. At that point it does 'unsetenv ("PATH");' and then tries to execute "echo". For this to work, we need to have echo in /bin, which we don't at that point. If /bin/echo -> /tools/bin/echo is added to the Essential Symlinks, I bet it will pass. Can you give that a try? Hmm. We may want to scour through the essential symlinks again. Greg seems to be getting away without grep and stty. http://www.diy-linux.org/x86-reference-build/chroot.html#c-create_symlinks > > That's one of those things I'd like to make > > cleaner/easier in jhalfs. > > Well, on normal failures the build dirs are keep. For forced failures like > this one, what I do is to not run automatically the Makefile, insert a "exit > 1" on the appropriate build script, and then launch manually the Makefile. > Very easy and quick, IMHO > > But if what you are thinking is on an option to keep all build dirs, that's > another beast ;-) That's what I had in mind. I have an idea of how to store a list of "keep_dirs" to compare against the current build/src directory, but I don't know if it will get really ugly. Manually killing it at the right spot works for now. -- Dan -- http://linuxfromscratch.org/mailman/listinfo/lfs-dev FAQ: http://www.linuxfromscratch.org/faq/ Unsubscribe: See the above information page