[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <4E21C317.2050306@bredband.net>
Date: Sat, 16 Jul 2011 18:57:59 +0200
From: magnum <rawsmooth@...dband.net>
To: john-dev@...ts.openwall.com
Subject: Re: LANG and LC_COLLATE
On 2011-07-16 16:33, Solar Designer wrote:
>> Putting LANG= before ./tstall will fix this. I enclose a patch for
>> Makefile that lets you run the test suite using "make test" from john
>> src directory, provided the test suite is installed in ../test relative
>> to that (i.e. test is a sibling to src). It includes this LANG= fix.
>
> Note that LC_COLLATE takes precedence over LANG, and I think LC_ALL does
> too. So if a system has LC_ALL or LC_COLLATE set, your fix won't help.
> I think the most specific one of these three - LC_COLLATE - is the higher
> priority one. So a better fix would be to set LC_COLLATE=C (not LANG).
Thanks, I did not know that. BTW, the best fix would be to edit tstone
in the test suite:
- sort < pw2 | uniq > pw3
+ sort < pw2 | LC_COLLATE=C uniq > pw3
Then it won't matter whether you run it directly or using "make test"
In my opinion this is a bug in uniq. National sorting order is one
thing, but who in their right mind would actually expect or *want* uniq
to filter out 'pass²' because there is a 'pass2'?
magnum
Powered by blists - more mailing lists
Confused about mailing lists and their use?
Read about mailing lists on Wikipedia
and check out these
guidelines on proper formatting of your messages.
