This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: Fix strtod rounding (bug 3479)
On Wed, 15 Aug 2012, Roland McGrath wrote:
> It's always helpful to push a private git branch to ease testing.
> Especially so when it's a large patch.
I don't think testing is a particular issue here; the only real
architecture-dependence for this and the previous patch is between 32-bit
and 64-bit architectures, and I tested both x86 and x86_64 to cover that.
strtod patches 3 (rounding mode support) and 4 (correct underflow
detection) would have more architecture-specific aspects, but I haven't
started work on those patches yet. The issue is more that the strtod code
is pretty complicated so patches aren't that easy to review. (A
significant proportion of the times I've studied the strtod code, I've
found new bugs in it.)
> That patch was still well under a megabyte uncompressed so I don't think it
> needed to be compressed for posting.
I was under the impression that there was a 400 kB message size limit, but
in fact it seems that while gcc-patches, binutils and gdb-patches all have
such a limit (400000 bytes), libc-alpha doesn't have a message size limit
at all.
--
Joseph S. Myers
joseph@codesourcery.com