==> On Thu, 09 Nov 2006 13:20:40 +0800, Ian Kent <[EMAIL PROTECTED]> said:

Ian> On Wed, 2006-11-08 at 22:37 -0500, Jeff Moyer wrote: > Hi, Ian,
Ian> list, (again, sorry if this is a dup)
Ian> > 
Ian> > Here is an implementation of included map support for autofs
Ian> v4.  This > supports included maps in both the master map, and
Ian> regular direct or > indirect file maps.
Ian> > 
Ian> > Because I dread writing bash code, I implemented the included
Ian> map > checking for the master map in C.  We basically just call
Ian> into the > daemon with a bogus mount point, asking for it to dump
Ian> out all of the > maps.  This works well for 2 reasons: 1) it's
Ian> not bash, and 2) it > re-uses the code that the daemon will use
Ian> to read the included maps > anyway.
Ian> > 
Ian> > Review would be greatly appreciated.  I tested this in my
Ian> environment > by creating a number of included maps, and
Ian> modifying > /etc/nsswitch.conf to ensure that it works as
Ian> expected for file and > NIS maps.

Ian> I've only looked briefly at this but I get the impression that it
Ian> will allow "+" map inclusion from other than file maps. Is that
Ian> right?

That was not the intent, and upon testing, it certainly doesn't follow
included maps from yp.  It does, however, interpret the included
map as a key in the map, which give a directory listing like this:

# ls /test/
+auto.test  repltest  sbox

Wrong, and somewhat amusing.  I'm not sure how best to deal with that.
I guess we need to throw out entries which do not have a "value"
portion.  I'm surprised we don't handle that already.

-Jeff

_______________________________________________
autofs mailing list
[email protected]
http://linux.kernel.org/mailman/listinfo/autofs

Reply via email to