Re: Building glibc 2.0.7 using hamm
On 29 Mar 1998, Ulrich Drepper wrote:
> Dale Scheetz <dwarf@polaris.net> writes:
>
> > Am I correct that when a cvs binary is installed, the config process
> > identifies it and configures cvs tests?
>
> This is all again much different.
>
> First, somebody who simply uses the tarball will never ever have any
> use for the CVS rules in the Makefiles. Only if a sources is changed
> some rules might trigger. Then it depends on whether the installer
> keeps the sources in a *private* CVS archive or not. If not, there
> are not CVS directories and nothing will happen. If there is a CVS
> directory, it points to her/his own CVS archive and an automatical
> checkin is ok.
>
> You miss probably that the CVS files are *not* part of the tarball.
>
I did remove the ./CVS node from the libc tree before I started making a
Debian package (I also added the necessary "add-on"s as per normal).
There doesn't seem to be any reference to the cvs archive, that this came
from, left in the system. The ./debian/rules file (the package building
make file for debian packages), does a make clean, then ./configure, and
then a make. The ./configure phase does some test calls to cvs, and since
it is installed on my system, notes that fact. Later during the build it
does some further tests which fail if the version is 1.9.10, but pass if
the installed version is 1.9.26.
I am willing to believe that the delivered tarball may have a different
confiugure, but the current one behaves this way. (Note: I've never
experienced this problem with glibc before, but I can't remember if I ever
had cvs installed previously as well)
Waiting is,
Dwarf
--
Still sigless
--
To UNSUBSCRIBE, email to debian-devel-request@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Reply to: