Bugzilla Bug 120403

sort +POS1 does not work anymore with cureutils-5.93. POSIX or bug?
Anyway, /etc/autofs/auto.net from autofs needs a patch:

--- auto.net.org        2005-11-08 19:49:54.000000000 +0100
+++ auto.net    2006-01-26 13:08:25.824136932 +0100
@@ -22,7 +22,7 @@
 # Newer distributions get this right
 SHOWMOUNT="/usr/sbin/showmount --no-headers -e $key"

-$SHOWMOUNT | LC_ALL=C sort +0 | \
+$SHOWMOUNT | LC_ALL=C sort | \
        awk -v key="$key" -v opts="$opts" -- '
        BEGIN           { ORS=""; first=1 }
                        { if (first) { print opts; first=0 }; print " \\\n\t"
$1, key ":" $1 }

most likely it's POSIX rearing it's "helpful" head

Created an attachment (id=78396) [details]
autofs patch

this patch is in fedora devel to fix sort in autofs

*** Bug 129642 has been marked as a duplicate of this bug. ***

*** Bug 129644 has been marked as a duplicate of this bug. ***

coreutils 5.93 has been stabled without this patch to autofs.  Most likely more
systems running "stable" will be broken now.  Please patch/bump autofs stable
to fix this.

the fedora patch (-k 1) is more appropriate than dropping the +0 argument
entirely (as in the original reporter's comment) as it results in the identical
behavior as +0 did.  Not that it REALLY matters what order the mount points are
processed in though :)

*** Bug 129780 has been marked as a duplicate of this bug. ***

*** Bug 130411 has been marked as a duplicate of this bug. ***

*** Bug 131859 has been marked as a duplicate of this bug. ***

(In reply to comment #5)
> coreutils 5.93 has been stabled without this patch to autofs.  Most likely more
> systems running "stable" will be broken now.  Please patch/bump autofs stable
> to fix this.
> 

This bug isn't moving. QA, please do something about this.

rphillips, robbat2: can one of you please add this patch and bump the latest
stable version?  If there is no response in a week, I'll assume it is alright
to add this, and I will bump it and add it.

halcy0n: you can go ahead and commit it.
I'm trying to work on the newer autofs stuff - which needs some major rewriting
of the init script again :-(.

Fixed.  Thanks Robin.

Bug is NOT fixed.  The added patch only "fixes" the auto.net file that is
installed in the samples.  The real problem is the auto.net file that is
installed from ${FILESDIR} into /etc/autofs/.  This file still contains the
sort +0.

Please change auto.net in ${FILESDIR}, or use the one from the sources and
rebump to really fix this annoying issue that still breaks stable systems.

Refixed

*** Bug 134741 has been marked as a duplicate of this bug. ***

this is really embarassing, but sorry... this fix again breaks a stable system:

% sort +k 1
sort: open failed: +k: No such file or directory

Not sure, but I think it should be "-k 1" which works fine.  Since I am no
member of any of the involved groups here, I don't like to commit a fix that
works... but please, could someone fix this for real very quickly now?  If you
need a patch reviewal on a system (or three) that uses autofs, please send me
mail, as I think it is clear that none of the committers actually could try the
script out before committing.

(In reply to comment #17)
> this is really embarassing, but sorry... this fix again breaks a stable system:
> 
> % sort +k 1
> sort: open failed: +k: No such file or directory
> 
> Not sure, but I think it should be "-k 1" which works fine. 


well, we had to wait only few months before seeing the patch available, but not
yet in the tree, with the huge change: sort -k 1

?????

(In reply to comment #17)
> this is really embarassing, but sorry... this fix again breaks a stable system:
> 
> % sort +k 1
> sort: open failed: +k: No such file or directory
> 
> Not sure, but I think it should be "-k 1" which works fine.  Since I am no
> member of any of the involved groups here, I don't like to commit a fix that
> works... but please, could someone fix this for real very quickly now?  If you
> need a patch reviewal on a system (or three) that uses autofs, please send me
> mail, as I think it is clear that none of the committers actually could try the
> script out before committing.
> 

No this was completely my fault, I even looked at the sort manpage before
changing it! :/

*** Bug 134929 has been marked as a duplicate of this bug. ***