Nils,
Don't know if you can help out, but not all platforms have the full
isnan/finite implementations. I can add the tests and remove these
fuctions for the platforms that don't have it, but I'm wondering if
it is possible to write the isnan and finite functions. For example,
on Windows, you have _isnan, but it is only good for doubles. There
is no float version. I looked at the macro for isnan on gcc and it
calls _isnand _isnanf and _isnan depending on the type size passed
in. But I don't want to start digging into gcc sources to see if I
could implement these three functions. Would you like to?
David
This is a multipart MIME message.
--==_Exmh_17931502740
cc:
Subject: IEEE Compute inquiries patch
Fcc: outbox
Mime-Version: 1.0
Content-Type: text/plain
--------
Enclosed is a suggested patch for making the isnan() and finite() functions
available in the Compute module. These functions accept TYPE_FLOAT and
TYPE_DOUBLE of CATEGORY_REAL as well as TYPE_FLOAT/CATEGORY_COMPLEX input
and returns an integer reflecting the outcome of the underlying call to
the isnan() and finite() functions of the math library.
The documentation update is rudimentary as I am not quite sure if the html
and help files are the right place to edit this kind of information or if
these pages are generated from some other master documentation.
This is mostly a cut-and-paste fix generated from the tanh and abs functions,
but it appears to work.
Cheers,
/Nils
--
.............................................................................
David L. Thompson Visualization and Imagery Solutions, Inc.
mailto:[EMAIL PROTECTED] 5515 Skyway Drive, Missoula, MT 59804
Phone : (406)257-8530
