Mail Archives: djgpp-workers/2003/09/07/15:11:25
Hello.
I've put a work-in-progress patch here:
http://www.phekda.freeserve.co.uk/richdawe/kbw-maths-20030907.diff.gz
It's half a megabyte uncompressed. I didn't think there was much point posting
it to djgpp-workers. I have no idea if it works. It compiles, at least.
You can find some notes I made here:
http://www.phekda.freeserve.co.uk/richdawe/kbw-maths-notes.txt
They're below for convenience.
K. B. Williams: I have a couple of questions:
* Could you send me a ZIP of your latest documentation for these functions,
please?
* Where are xexp2, xexp1m defined?
Thanks, bye, Rich =]
Integration of K. B. Williams's new math functions
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
* C99 header guards.
- Merging math.h manually piece-wise (function by function), so OK.
* fpclassifyl, isfinitel, isinfl, isnanl not integrated.
- Use MartinS's instead.
* nanl - go via a __dj_* like nan we added for C99?
- Need to add nan*(const char *) functions. Simple wrappers
that invoke strto*.
* float.h: Don't put the new constants there, since they're not defined
in C99. Put them in a private header instead in src/libc/c99/math.
* tests/libclink: Missing checks for C99 functions? modfl
* signgaml should be prefixed with underscores, so it doesn't pollute
the namespace.
* Fix tgen builds: gcbrtvec.cpp, gexpm1vec.cpp.
- Where is xexpm1 defined?
* Fix tgen builds: gcosvec.cpp, gexpvec.cpp, gsinvec.cpp.
- Where is xexp2 defined?
- Other x* functions appear to be missing too.
* Fix tgen builds: gtanhvec.cpp.
- _nanf.
* math.h, libm/math.h - Finish integrating KBW's diffs. How to make
libm/math.h just a shell without any fdlibm pollution?
- Sort out fdlibm.h & math.h clashes properly.
Richard Dawe <rich AT phekda DOT freeserve DOT co DOT uk> 2003-09-07
--
Richard Dawe [ http://www.phekda.freeserve.co.uk/richdawe/ ]
- Raw text -