Originally Posted by rworkman
If you modified the script to say "CLANG=yes" then you shouldn't need to re-run it.
Did you actually install the resulting package after it built?
So I still have to invoke pkgtool or installpkg on the generated tarball in the /tmp directory?
I presumed that the script ran a make install (or equivalent) step at the end.
I have since tried to build version 3.0 by downloading the source tarball from the Clang/LLVM website and even after altering the VERSION=2.9 to VERSION=3.0 in the script and changing the name of the actual source tarball slightly (it was named llvm-3.0.tar.gz when the script expected llvm-3.0.tgz) the script hangs up after initially starting and running briefly. I presume this is secondary to the tarball being arranged in a way it did not expect. The LLVM team originally implied that the Clang source be placed in a llvm/tools directory within the extracted llvm source while the Slackware Builds seems to place both as gzipped tarballs in the same directory as the extracted SlackBuilds script. Just to clarify this attempt to build version 3.0 is a 3rd attempt after having run (and rerun) the build on the 2.9 source tarball downloded via the link in SlackBuilds.
Is there a particular trick to trying to be on the bleeding edge with some Slackware builds? Should I be fetching the source tarballs from a particular location? Or just wait for the sources to be linked to from SourceBuilds. What I am asking is how tied am I to running the version stated in the SlackBuilds site?
The SlackBuilds script seems to create one tarball that presumably has both LLVM and Clang. Will test it later. I have not "installed" yet if I understood you correctly.
Sorry for my stream of questions. I am learning a lot in the process.