This is the mail archive of the
gsl-discuss@sourceware.org
mailing list for the GSL project.
Re: [Bug-gsl] Potential Bugs?
- From: Brian Gough <bjg at gnu dot org>
- To: Thanh Vo <vo at ucdavis dot edu>
- Cc: gsl-discuss at sourceware dot org,bug-gsl at gnu dot org,"Earl T. Barr" <etbarr at ucdavis dot edu>
- Date: Tue, 17 Aug 2010 21:47:29 +0100
- Subject: Re: [Bug-gsl] Potential Bugs?
- References: <AANLkTinA0Bo1tS-AuqeE62O=SqcDSMffHWKbxh5G4a1V@mail.gmail.com><43mxsl6klg.wl%bjg@gnu.org>
At Tue, 17 Aug 2010 21:38:19 +0100,
Brian Gough wrote:
> The overflow exception is a valid warning. It's also a bug in the
> sense that it is avoidable (although the end result is correct, z=-1
> since 1/inf=0). It would be better written as (32/y)/y.
Of course, apart from the extra factor of 2^5 that gives an underflow
instead but it's probably preferable to an overflow.