[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Is the FHS dead ?

Josselin Mouette wrote:
Le mardi 17 février 2009 à 07:41 +0100, Giacomo Catenazzi a écrit :
Also for cgroups, I really hope that proposal come from distributions
(and common usage). Only after one distribution use it, it need to
be standardized.  IMHO standards should come from bottom.

I’m not saying that something like cgroups should be directly written in
stone to the standard. However the discussion should have taken place at
a cross-distribution level so that it could be standardized later, if

Which btw it was done in debian-devel/linux-kernel, involving the major distributions

Anyway, I think that we must give new live to fhs mailing list
(also for discussion, not only for proposal).

Do you really want to subscribe to a list that only ships 50 spams/month

ok. This is a huge problem. Maybe you should ask to the kernel mailing list +
the FHS maintainers, to move the list in kernel.org or other less spam attractive

BTW I had also other though about the issue, and I think we need a
"Debian FHS", which should reference the main FHS, but it should
precise better some issues:
- Debian specific location for package management (houside scope of FHS)
- stricter rules, where FHS is too liberal (cgi, python,...)
- specify stricter rules about attributes:
   - temporary partitions (see recent debian-policy discussion about
     /var/run on tmpfs)
   - which zone could be: ro, nodev, nosetuid, noexec
   - which directories should not be symlinks
     (there are some policy requirement about relative symlink)
- transitions: package maintainers IMHO should still allow
   old directories (but not create new think in wrong places),
- things not yet in FHS

I think centralizing such document would be nice.


Reply to: