On Sat, May 14, 2011 at 11:34:57PM +0000, Christoph Anton Mitterer wrote: > 5) Emphasis that "/mnt" is for a _single_ _temporary_ mountpoint...i.e.: > - nothing that should go to /etc/fstab > - not having subdirectories (does cgconfig/cgred in Linux still use this?)
FHS 2.3 already states that the contents of /mnt is a local issue. Therefore, nothing FHS complaint should rely on anything being anywhere inside /mnt. Why make this stronger? > 6) Relax unnecessarily strict requirements: > "/sbin" requires to have all fsck.* and mkfs.* tools. > - IMHO it neither makes sense to restrict them to /sbin (but also allow > /bin, because nowadays there may be filesystems that are user-centric (and > not device-centric),... e.g. on could think of things like mkfs.gmail, > which automatically registers a google account and mounts it in Linux via > FUSE. FUSE is an interesting case. I think such "user filesystems" are not under the scope of the FHS, and where it mandates mkfs.* for an installed *subsystem* at present, that should be system-wide filesystem subsystem -- perhaps the text needs clarification. > - It also does not make to keep it away from the "/usr/bin" or "/usr/sbin"... > as many of those filesystems are in no way required to boot or to run the > system. What filesystems are necessary to boot the system are largely dependent on local decision-making. How can the FHS know whether you need ext3 or reiser4? At best it might know that you don't need gmailfs (see above) -- Jon Dowland _______________________________________________ fhs-discuss mailing list [email protected] https://lists.linux-foundation.org/mailman/listinfo/fhs-discuss
