On May 4, 2012, at 4:05 AM, Pinto Elia wrote: > > CLEANFILES definition had a typo >
Thank you! Yes this was the easy fix (that I wasn't seeing). Still the comment here is indicative that the entire approach @rpm.org is/was horrendous overkill # Modified to use sha1sum instead of cmp to avoid a diffutils dependency. So all of this implementation was tortured to "prefer" sha512 simply to confirm to the dreadfully uninformed Fedorable Packaging Polizia. Avoiding a dependency on diffutils MIGHT have been resolved by putting cmp(1) into its own package: diff(1) isn't exactly "bloat" either. (aside) Fascinating (to me) seeing how package monkeys adapt to externally imposed constraints. Its kinda like cheese in a maze or sugar in an ant farm, where swarms of soldiers struggle to solve simple problems "correctly". Meanwhile its likely time to attempt a general solution to hardlink identical content both before packaging in rpmbuild, and during installation in rpmlib. The needed file digests to identify identical content have been in place for years, the RFE from Russ exists, and the only remaining implementation need would be to be able to disable the mechanism on a per-path basis so that sysadmin's editing a hardlinked file don't accidentally alter initially identical content without creating a copy and breaking the hard link. 73 de Jeff ______________________________________________________________________ RPM Package Manager http://rpm5.org Developer Communication List rpm-devel@rpm5.org