On Fri, 2007-11-23 at 01:55 -0500, Tom Lane wrote: > Ralf Corsepius <rc040203@xxxxxxxxxx> writes: > > On Thu, 2007-11-22 at 22:03 -0800, Peter Gordon wrote: > >> Incorrect. Invoking autotools modifies the package's build scripts, not > >> source code. There is a significant difference. > > > With all due respect, you could not be much wronger. > > > auto* generated files are not supposed to be dynamically modified nor > > are they designed to be dynamically modified. > > That's the reason why they are part of source tarballs. > > No, the reason why they are part of source tarballs is that > traditionally tarball-makers haven't wanted to assume that users will > have autotools installed. Right, that's one aspect. It's one aspect why they are treated as generated SOURCES. > > From a practical aspect: > > Running incompatibles version of the autotools during builds can break > > a package. > > Certainly true. On the other hand, *not* rebuilding the scripts can > also break a package, particularly with respect to libtool, IME. Libtool is a special case, because RH ships a hacked libtool to work around an issue (multilibs) upstream so far has not been able to fix. > It's a case-by-case issue, and so I see any sort of generalized > opinionating about it as fundamentally misguided. Wrong. Read the Fine Manual, read the GNU Coding Standards. > I do a forced > automake/libtoolize/autoconf for some of my packages, and not others. Your mistake. You are unnecessarily imposing risks to your packages. Ralf -- fedora-devel-list mailing list fedora-devel-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/fedora-devel-list
