ever since I took a FreeBSD machine into production, acting as any kind
of file server, I have to work arround the problem, that write access to
a directory implies unlinking (deleting) directory contents. Never heard
any sensible explanation why anybody would ever want that behaviour, but
it's been like that for decades and everybody seems to be fine with
that!?! Maybe because there's the stick bit, which is a usable workarround.
Unfortunately, there's no “sticky” equivalent in nfs4acls.
More unfortunate, newly created directories don't inherit the sticky bit
– unlike the group settings.
And most unfortunate, I'm not able to implement sticky bit inheritance
myself :-(

Since there's already a kind of inheritance when calling mkdir(1), I
guess extendig the inheritance to respect the sticky bit shouldn't be
too complex, is it?
I'd love to see a sysctl which controls the behaviour, so there's no
unexpected behaviour, but the possibillity to make FreeBSDs
filsystem-permission-control more real-world-usable. But if a sysctl is
noticable more effort than just a kern-conf (compile time) option, I'd
also highly appreciate that option!

Is there anybody who might want to look into that?



Attachment: signature.asc
Description: OpenPGP digital signature

Reply via email to