[Bug libc/14049] Incorrect strtod, atof etc. hex float rounding

["joseph at codesourcery dot com" <sourceware-bugzilla at sourceware dot org>]

http://sourceware.org/bugzilla/show_bug.cgi?id=14049

--- Comment #2 from joseph at codesourcery dot com <joseph at codesourcery dot com> 2012-05-02 19:23:21 UTC ---
Rounding according to the current rounding mode is not part of this issue, 
which is about one particular bug in the code addressed by my patch 
<http://sourceware.org/ml/libc-alpha/2012-05/msg00076.html>, or of bug 
3479 either; it's a separate issue (maybe one without an open bug at 
present).  In most cases, bugs should cover one limited well-defined 
issue, and patches should be incremental and fix the least unit it makes 
sense to fix in one patch, so that they are easier to produce and easier 
to review (and so a fix for one issue can get in quicker even if there are 
problems with the fix for another related issue).  It's natural to fix the 
bugs with round-to-nearest rounding (hex and decimal separately) before 
doing anything about current-rounding-mode rounding.

As discussed for printf in bug 5044, using fegetround would have the 
problem that it's only in libm not in libc; although it could be added to 
the set of libm functions duplicated in libc, fpu_control.h may be better 
but it doesn't have all the relevant macros on all architectures 
supporting rounding modes and there's a need for architecture-specific 
support for getting rounding modes for cases such as ARM and Power 
soft-float.

-- 
Configure bugmail: http://sourceware.org/bugzilla/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.