The NaN and Inf catching functions 'isnan()' and 'isinf()'.The NaN and Inf catching functions 'isnan()' and 'isinf()'.Hi,
I have merged the 'nan_catch_test' branch back into the 1.2 line. The commit message of the merge is located at https://mail.gna.org/public/relax-commits/2006-09/msg00008.html (Message-id: <E1GJnLY-0000on-Dj@xxxxxxxxxxxxxxxxxx>). The new code attempts to fix bug #6503 (https://gna.org/bugs/?func=detailitem&item_id=6503) and bug #6510 (https://gna.org/bugs/?func=detailitem&item_id=6510). I've closed bug #6510. It implements a number of ideas discussed in the threads starting at https://mail.gna.org/public/relax-devel/2006-08/msg00000.html (Message-id: <7f080ed10608020558o47fd0260s9579f663147ab84@xxxxxxxxxxxxxx>) and https://mail.gna.org/public/relax-devel/2006-08/msg00029.html (Message-id: <7f080ed10608082240r114cf35an60af62836fcc81bf@xxxxxxxxxxxxxx>).
I have added two functions called 'isnan()' and 'isinf()' within the file 'float.py' in the base directory. These, however, are currently dummy functions and the minimisation code in 'specific_fns/model_free.py' which calls them has been commented out. We still need to implement the bit pattern matching (including the handling of 32 bit vs. 64 bit, big-endian vs. little-endian, etc). Another good link is http://babbage.cs.qc.edu/courses/cs341/IEEE-754references.html. Work can continue on these functions without disrupting normal development. I have also created task #3621for this job which is located at https://gna.org/task/?func=detailitem&item_id=3621.
Oh, according to PEP 754 (http://www.python.org/dev/peps/pep-0754/) the 'fpconst' is trying to get into the Python standard library. It should hopefully be included in the new 2.5 version of Python (http://www.python.org/download/releases/2.5/). They've just released rc1 of that code.
Edward