Le dimanche 11 mars 2007 à 01:12 +0100, Axel Thimm a écrit : > On Sat, Mar 10, 2007 at 05:32:03PM +0100, Nicolas Mailhot wrote: > > Please can we kill the "fixed UID are useless", [...] > > > The *only* problem is this range is full, not that it should or should > > not exist. Any other argument is not worth the bytes expended on it. > Maybe some quick review will give us 42 *fixed* uids, and I'm > sure that's more than enough for the next 1-2 years (e.g. until > F9/10/RHEL6. And until then we can talk with the LSB to change the > system fixed/non-fixed uid ranges and prepare a sensitive and > compliant setup to last for the next decade. IMHO trying to be smarter at fixed UID attribution is a dead end. We've been hitting the limit for some years now, and getting smart just didn't happen. It won't now — it's just too much work to evaluate the threshold between fixed/dynamic (taking into account past of future versions of a package, build and configure options, local site usages, etc) A KISS policy "every rpm-created system ID has a fixed uid referenced in this table" is the only sane approach. Except for the problem of short-sighted range limit, UIDs are cheap and not worth spending hours over. -- Nicolas Mailhot
Attachment:
signature.asc
Description: Ceci est une partie de message =?ISO-8859-1?Q?num=E9riquement?= =?ISO-8859-1?Q?_sign=E9e?=
-- Fedora-maintainers mailing list Fedora-maintainers@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/fedora-maintainers
-- Fedora-maintainers-readonly mailing list Fedora-maintainers-readonly@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/fedora-maintainers-readonly
