At 11:24 Uhr +1100 09.02.2002, [EMAIL PROTECTED] wrote: >On Saturday, February 9, 2002, at 10:45 AM, Peter O'Gorman wrote: > >>> >>>That won't work with a shared fink folder. I pointed that out in >>>my original mail I believe :) >>Yes, I really ought to learn to read... >> >>Fink could write to /sw/var/lib/dpkg/available directly (this isn't >>going to be shared). >>Changing dpkg doesn't seem like the "right thing". > > >Changing the deb package database formats would probably be a bad idea.
Agreed. >Changing dpkg to have a system wide set of preinst postinst, and >prerm scripts (Perhaps in a directory run with run-parts) that would >be called at sensible times. It may even be accepted in the upstream >sources, however I think they have other things to worry about like >releasing debian 3.0. Hu, why?!? I dunno how you suddenly jumped to this topic? > >How is this for an idea for what a darwin-5.2 psuedopackage might do. > >Place a script in the global preinst script.d dir that does >something like this. That's not very logical. Dpkg already has a built in sytem to handle versionened dependencies. If we are going to modify dpkg, then we should base our changes on that existing mechanism, instead of abusing an install script for this. And that's what it is, an abuse. A proper dependency field will immediatly tell the user he can't install package; but the installscripts you and david suggest will fail *during the install*. Not good. So far I haven't seeny any proposal I like, sadly (that includes mine). Hrmpf. Max -- ----------------------------------------------- Max Horn Software Developer email: <mailto:[EMAIL PROTECTED]> phone: (+49) 6151-494890 _______________________________________________ Fink-devel mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/fink-devel
