This is the mail archive of the
fortran@gcc.gnu.org
mailing list for the GNU Fortran project.
Re: [Patch, wwwdocs] gcc-4.3/changes.html: Fortran - Add GAMMA/LGAMMA
- From: "Smith-Rowland, Edward M" <ESmith-rowland at alionscience dot com>
- To: <fortran at gcc dot gnu dot org>, <tkoenig at netcologne dot de>
- Date: Wed, 31 Oct 2007 16:22:03 -0400
- Subject: Re: [Patch, wwwdocs] gcc-4.3/changes.html: Fortran - Add GAMMA/LGAMMA
------------------------------------------------------------------------
------
I am currenly working on fallback implementations of tgamma*() and
lgamma*() using the Lanczos approximation (one reason why I've been
quiet these last days :-) I've gotten pretty far by now, but you
can expect a patch within the next few days.
I hope we didn't do too much duplicate effort here.
Thomas
------------------------------------------------------------------------
------
I don't know how much this will help but I did a similar thing for
libstdc++ a while back.
This is in /trunk/libstdc++-v3/include/tr1/gamma.tcc on SVN.
http://gcc.gnu.org/viewcvs/trunk/libstdc%2B%2B-v3/include/tr1/gamma.tcc?
revision=125244&view=markup
The implementation is pretty templatized (so we could support all the
floating types) so translation might be a little annoying.
Also, in C++, we didn't touch complex because it's pretty tough in
general.
This helped our implementation of C++ TR1 math special functions to work
on systems without a math processor or anything.
There are other tweaks to the Lanczos algorithm that one could use
(different parameters). I'm interested to see what you come up with.
Ed Smith-Rowland