X-Authentication-Warning: delorie.com: mail set sender to djgpp-workers-bounces using -f
Sender: rich AT phekda DOT freeserve DOT co DOT uk
Message-ID: <3FA164AB.EC4594A8@phekda.freeserve.co.uk>
Date: Thu, 30 Oct 2003 19:21:15 +0000
From: Richard Dawe <rich AT phekda DOT freeserve DOT co DOT uk>
X-Mailer: Mozilla 4.77 [en] (X11; U; Linux 2.2.23 i586)
X-Accept-Language: de,fr
MIME-Version: 1.0
To: djgpp-workers AT delorie DOT com
Subject: Re: locale fixes
References: <3F9E3836 DOT 000008 DOT 02793 AT soapbox DOT yandex DOT ru> <3F9EC775 DOT 41795CD0 AT phekda DOT freeserve DOT co DOT uk> <3F9FD9C8 DOT 000003 DOT 05203 AT ariel DOT yandex DOT ru>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Reply-To: djgpp-workers AT delorie DOT com

Hello.

Alexander Aganichev wrote:
[snip]
> >I don't have time to review them properly, but they basically look OK to
> >me. I'll review them properly on Sunday, hopefully.
> >
> >Do you have any test programs for these functions?
> >
> >If there are any existing tests, do they still pass, after applying your
> >patches?
> 
> In general, without new setlocale() all locale patches should do nothing
> more then all those functions currently do. They only add the possibility
> to modify the behaviour of these functions from external modules.

Sure, it should work the same. ;) But how can you prove it without some
regression tests?

> For decimal point fixes I'll write the test cases where decimal_point is set
> from the application to some odd character (like comma or even ampersend)
> and test that conversion from ASCII and to ASCII still stable.

Sounds good.

> For strcoll and strftime it is possible to modify static data and test that
> the result is also changed. I tested it with new setlocale() on Russian
> locale.

Sounds good.

> I'll post these test cases before the end of this week.

Thanks. That would be appreciated.

> I have no good idea how to check fix for the realloc() in regex. I did it
> by the visual examination of code.

I was thinking of something like this:

Write some regression tests for regexps. (Actually, there are already some.)
Modify realloc so that it fails randomly. Rebuild libc and test. Check that
nothing bad happens.

Bye, Rich =]

-- 
Richard Dawe [ http://www.phekda.freeserve.co.uk/richdawe/ ]

"You can't evaluate a man by logic alone." -- McCoy, "I, Mudd", Star Trek