If there is any missing functionality that is required by the C99 standard, that's an important bug, and we should file it separately.
I don't think the standard requires all functionality to be available as function pointers. So you'll probably need to keep your wrapper functions. Rather than writing your code to know which functions are available as function pointers on which platforms, wouldn't it be better to write the code so that it runs on any C99 compliant system? If you want to request direct function pointers for performance reasons, then you should be explicit about the performance difference, and explain the positive impact on ksh users that would result from improving the performance. --chris ольга крыжановская wrote: > Thank you for filing the bug. > > I see it has been closed as will not fix. Why? AFAIK all C99 functions > must be available as float, double and long double functions, or not? > > Olga > > 2010/5/24 Chris Quenelle <chris.quene...@oracle.com>: >> If you are requesting general additions to libm, that sounds like it's >> almost completely unrelated to phasing out libsunmath. >> It would be helpful to start a new email thread, and introduce >> your new topic with a summary. >> >> The email that you responded to was asking for a set of >> libsunmath functions that existing libsunmath users found >> useful. It was not a general request for libm RFEs. >> >> I have filed a libm RFE for you. >> 6955163 request for additional libm functions >> >> It should be visible on bugs.sun.com within a few days. >> >> --chris >> >> >> >> ольга крыжановская wrote: >>> Useful functions would be long double versions of the existing C99 >>> functions. We (the ksh93 integration team) found that several >>> functions are missing in libm and require to add substitute wrappers. >>> Using libsunmath is unfortunately not an option because it breaks the >>> gcc build of ON. >>> Currently these function are either not available in libm, not >>> available in versions taking long double as argument or not available >>> as function pointers: >>> finite >>> fpclassify >>> isfinite >>> isgreater >>> isgreaterequal >>> isinf >>> isless >>> islessequal >>> islessgreater >>> isnormal >>> issubnormal >>> isunordered >>> iszero >>> signbit >>> >>> Olga >>> >>> 2009/10/23 Chris Quenelle <chris.quene...@sun.com>: >>>> ольга крыжановская wrote: >>>>> Can you merge libsunmath into libm? >>>>> >>>> If we could identify the useful functions, we could >>>> move those into libm. Or some other more appropriate library. >>>> >>>> --chris >>>> >>> >>> >> > > > _______________________________________________ ksh93-integration-discuss mailing list ksh93-integration-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/ksh93-integration-discuss
