Daniel P. Berrange wrote:
I think the source code should go into gnulib/*.[ch] in case we ever want to have a lib/ dir for code shared by the daemon & library. There's no need to pollute the top level dir with gl-tests, when we can have tests/gnulib/, or gnulib/tests/. We've already got an m4/ directory, so we might as well use that (or a m4/gnulib, or gnulib/m4 subdir).
Minor point here, but the way that Jim's bootstrap script is set up, it will create a gnulib/ temporary directory if it needs to git-clone gnulib.
I don't want a dependancy on a script pulling in files from another random project that we have to grab from the internet.
And this, I think, is the reason to check in the generated source files to CVS. It's not something I'd normally advocate, but there is surely a special exception to be made if (a) the file has to be downloaded from somewhere or (b) the file rarely changes and is hard to generate. The 'rpcgen'-generated files fall into (b), and the gnulib files fall into both categories.
Given a fresh CVS checkout of libvirt & fresh Fedora/RHEL/etc OS install one should be able to do a full build without requiring extracode checkouts from unrelated projects. Since gnulib is only distributed in source form, and not available from the OS package manager, the only viable approach is to commit the GNULib files to CVS.
Agreed. Rich. -- Emerging Technologies, Red Hat - http://et.redhat.com/~rjones/ Registered Address: Red Hat UK Ltd, Amberley Place, 107-111 Peascod Street, Windsor, Berkshire, SL4 1TE, United Kingdom. Registered in England and Wales under Company Registration No. 03798903
Description: S/MIME Cryptographic Signature