-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Ben Reser wrote: | On Thu, Nov 14, 2002 at 10:38:05AM -0500, R P Herrold wrote: | |>1. The 'make install' decision is that of the initial package |>maintainer, not of RPM -- if you don't want the documents |>installed, exit the makefile or Makefile.in with a patch file | | | This simply replaces one problem for another. Patching makefiles is | bound to create other bugs. It's silly to have to patch a makefile just | to leave out an unneeded file.
No, it's forcing the lazy developer (packager) to take an active stance. You sent an email complaining how no one tests packages. Well, this is a part of that. It forces them to not be lazy and exclude certain files because they just forgot to add them to the files list. One drawback I see, however, is that the lazy developer will again become lazy and start using wildcards all over the place in the filelists. That's a bad idea too. | That's better than modifying the makefile. But it's still stupid. | I can think of many reasons why files might not get included that a | package might install. Things from Mandrake not shipping the matrix | screensaver to a package installing it's documentation in the wrong | place. rpm has no way of knowing if you're doing something smart or | something stupid. Ultimately it takes a human making the correct | decision. | | Forcing us to rm everything is a waste of time. Patching the Makefile is out of the question, but I also agree that removing many files is a hassle. But what is your proposed solution? | | |>3. Failing that, at RPM package install time, any package can |>be installed without items identified in a %doc stanza -- |>there is a --nodocs option on install, and that will keep them |>out of an installed system. | | | Huh? This is not a solution to the problem at hand. It's a nice option | for users but does nothing for packagers. And it certainly does nothing | for issues above and beyond documentation files. Actually, it creates a problem for packagers since rpm will see these files as missing, whereas they were decidedly not installed, rather than missing or deleted. | This does nothing to fix the tons of issues of people not following | Mandrake guidelines. rpm will still blissfully build packages with bad | menu systems. It still will build rpms that don't follow the allowed | Groups. It still lets people use xpm's instead of pngs. It doesn't | complain about gzip instead of bzip2. etc etc etc. Yeah, but we know the solution here, and it's this: Integrate rpmlint into one of the rpm build scripts. If rpmlint fails, abort the package creation. We need to take an active stance, Mandrake employees aren't even doing it and they are getting paid. You can imagine that an unpaid voluteer is even lazier. Granted, I volunteer, but I am also passionate about anything I do. I try to take this active stance we're talking about, but some people need to be forced/have there hands held. Plus, it's easier to deny a volunteer contribs access (heck I don't even have contribs access), than it is to fire a MandrakeSoft employee who has always made sloppy packages. The community doesn't decide either directly or indirectly about people's employment. - -- Sincerely, David Walluck <[EMAIL PROTECTED]> -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.1 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQE90/6oJB7s/tlHKVIRApKqAJ4qy2db/+ljqoZQN/v3IQhUtpXYLwCfeGLb K18mkQhFWkVnssruhY48XLg= =JUAK -----END PGP SIGNATURE-----
