Re: [Gcl-devel] ANSI test fixes

[Camm Maguire]

Greetings!

Looking at this briefly, I don't think the recent changes are
related.  It does not appear that axiom really wants ash here in the
negative case.  GCL's ash (with one small exception just fixed)
concurs with the ash of the other ansi compliant lisps to my
understanding. 

The recent changes primarily relate to float type promotion problems
certain functions were showing, plus the repair of a problem not in
2.6.5 that was recently introduced into 2.7.0 when accelerating ratio
operations. 

Take care,

"Page, Bill" <address@hidden> writes:

> Camm,
> 
> Do these fixes for round and truncate etc. have anything to
> do with Axiom bug #4733 and the patch that I proposed some
> time ago for that?
> 
> http://savannah.nongnu.org/patch/download.php?item_id=2074&item_file_id=2073
> 
> I notice that this problem still occurs in the version of
> Axiom compiled from June 21, 2004 sources.
> 
> Cheers,
> Bill Page.
> 
> > -----Original Message-----
> > From: Camm Maguire [mailto:address@hidden
> > Sent: Tuesday, August 24, 2004 11:16 PM
> > To: address@hidden
> > Subject: [Gcl-devel] ANSI test fixes
> > 
> > 
> > Greetings!  Just committed some corrections and cleanups to the recent
> > accelerated ratio algorithms, and fixes for the (f)floor, (f)ceiling,
> > (f)round, (f)truncate, atan, asin, acos, and sqrt ansi tests.
> > 
> > Please test and enjoy.  Feedback of course always appreciated.
> > 
> > Still to do is to allow inlining of transcendentals from libm under
> > appropriate circumstances.
> > 
> > Take care,
> > 
> > -- 
> > Camm Maguire                                                
> > address@hidden
> > 
> 
> 
> 

-- 
Camm Maguire                                            address@hidden
==========================================================================
"The earth is but one country, and mankind its citizens."  --  Baha'u'llah