Sender: rich AT phekda DOT freeserve DOT co DOT uk Message-ID: <3D02463F.B5A4C338@phekda.freeserve.co.uk> Date: Sat, 08 Jun 2002 19:00:31 +0100 From: Richard Dawe X-Mailer: Mozilla 4.77 [en] (X11; U; Linux 2.2.19 i586) X-Accept-Language: de,fr MIME-Version: 1.0 To: djgpp-workers AT delorie DOT com Subject: Re: Patches for building with gcc 3.1 - *scanf() chunk References: Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Reply-To: djgpp-workers AT delorie DOT com Hello. Eli Zaretskii wrote: > > On Wed, 29 May 2002, Richard Dawe wrote: > > > Please find below a patch for building with gcc 3.1. This converts > > _doscan*() to using va_list. This seems to work fine with programs > > in the test suite. > > The patches look okay by inspection, but please be sure to run the Cygnus > test suite (from djtst203.zip) before and after the change, and closely > scrutinize any differences in the part that tests formatted I/O routines. > Changes in math results can also be due to reading/writing numbers by > printf and scanf families, so please look at that as well. I've compared the following check-outs: a) DJGPP CVS sometime ago - this does not have any of the gcc 3.1-related commits; b) DJGPP CVS updated to latest; c) DJGPP CVS with all my patches. I rebuilt libm for all the cases with gcc 3.1. Then I ran the test suite. They all give the same mtest.results file. It can be found here: http://www.phekda.freeserve.co.uk/richdawe/djgpp/misc/mtest.results However, more tests fail than in the std-*.results files in tests/cygnus in djtst203.zip: File Failures ---- --- std-v202.results from djtst203.zip 395 std-libc.results from djtst203.zip 478 mtests.results from a), b) or c) 525 I don't think the *scanf() patch has caused an increase in the number of failures, since the result is the same with and without the patch. So, if there are no objections, I will commit the patch after waiting the usual week. Bye, Rich =] -- Richard Dawe [ http://www.phekda.freeserve.co.uk/richdawe/ ]