On Thu, Jul 03, 2003 at 04:23:03AM -0400, David B Harris wrote: > On Wed, 2 Jul 2003 13:15:09 -0500 > Steve Langasek <vorlon@netexpress.net> wrote: > > Section 10.2 of policy currently describes uid and gid classes covering > > the range of 0-65535. This appears to no longer be comprehensive: on a > > current system running a 2.4.18 kernel and libc6 2.3.1-17, I'm able to > > assign 32-bit userids to accounts and reference these accounts in file > > ownerships, su to them, etc. Should Debian Policy be expanded to > > address this greatly increased range of available ids? > I certainly agree with the general idea, as well as the specific > proposal of allocating 2^16 UIDs for Samba's idmap usage. > That being said, will Sarge release with the minimum requisites for the > 2^32 UIDs? If so, I'm happy. But somebody should ask the RM to be sure. > Otherwise, I would think it'd have to wait until Sarge+1. > Specifically, quota stuff. Every tree but Marcello's has implemented 32 > bit quotas, as far as I know, but not his. So 32 bit quotas aren't > "official" yet. Might need Xu to patch the default kernel images. If Debian ever hopes to have a policy beyond "all remaining uids and gids are reserved for local use", it's important to stake our claim *before* 32-bit ids are universally supported -- that is, before they're in widespread use at sites, and site admins have already deployed schemas that conflict with any default we might choose. I'm content if the 32-bit id support in sarge is not a 100% solution. Defaults never work for everybody. But the way things are looking now, the current defaults (no available range) will work for a diminishing proportion of users. I don't see how a default uid range that some people can't use is worse than an empty default uid range that no one can use. -- Steve Langasek postmodern programmer
Attachment:
pgpUEFkSsxeIC.pgp
Description: PGP signature