bug#18540: Sorting bug?

[Eric Blake]

tag 18540 notabug
thanks

On 09/23/2014 02:58 PM, Eric Blake wrote:

> Let's look further:
> 

> $ printf 'a b!x\na-b-c!x\n' | LANG=en_US.utf8 ltrace -e strcoll sort -s
> --debug -k1,1 -t!
> sort: using ‘en_US.utf8’ sorting rules
> sort->strcoll("a b!x", "a-b-c!x")                = 21
> a-b-c!x
> _____
> a b!x
> ___
> +++ exited (status 0) +++

Hmm, I just noticed something.

> 
> 
> Huh? Why are we passing the ENTIRE line to strcoll?  Shouldn't we only
> be passing the key?

That was my distro's build of sort (in my case, Fedora 20, with sort
from GNU coreutils 8.21).  But looking at coreutils.git (v8.23-39-g1ff4d08),

$ printf 'a b!x\na-b-c!x\n' | LANG=en_US.utf8 ltrace -e strcoll
./src/sort -s --debug -k1,1 -t!
./src/sort: using ‘en_US.utf8’ sorting rules
sort->strcoll("a b", "a-b-c")                    = -1
a b!x
___
a-b-c!x
_____
+++ exited (status 0) +++

Yay - strcoll now uses the correct bounds.  Next step - determining if
this is an upstream problem that was fixed in the interim, or if this is
a bug in the downstream additions on top of stock upstream.  None of the
9 commits in 'git shortlog v8.21.. src/sort.c' seem to describe the
situation.

And looking at my distro's patches, there is definitely some gorp added
to sort.c in coreutils-i18n.patch, which I highly suspect to be the root
cause.

So please re-raise this as a downstream bug in your distro's i18n patch,
as upstream coreutils is immune.

-- 
Eric Blake   eblake redhat com    +1-919-301-3266
Libvirt virtualization library http://libvirt.org

signature.asc
Description: OpenPGP digital signature