Hi ! I finally understood what to do by carefully looking at FloatMgr.h and SampleCalc.c files comments ( always RTFM twice !! :) ( maybe the prc doc or FAQ could be a good place... ) So, it seems that with GCC, we can use the new floatting point manager, but not the new functions ( _d_mul and so on ), even with the libnfm.a linked with the program. As english is not my natural language ( but I'm sure you already figured that yourself :), it was not clear to me the first time I read all the docs. Maybe a good thing, to avoid these kind of problems, would be to generate an error if we use any of these functions ? Anyway, as mentionned in SampleCalc comments, I have replaced the calls to _d_mul and so on functions with the direct operator ones ( I don't think I need FlpBuffer* functions for my problem ). To be sure what I've done is right ( I'm not familiar at all with FP ), here is a code snippet : FlpCompDouble a, b, c, v; Int32 val; Boolean result; (...) Old version : v.fd = _d_itod(val); a.fd = _d_mul( v.fd, b.fd ); result = _d_cmp( a.fd, c.fd ) == flpLess; New version : v.d = val; a.d = v.d * b.d; result = ( a.d == c.d ); Do the FP gourous think this is fine ? Is the 'direct' conversion from Int32 to Double ok, and did the test to compare the result do the same thing than _d_cmp with flpLess ? Thanks. -- Daniel Morais -- For information on using the Palm Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/tech/support/forums/
