This is the mail archive of the
gdb-patches@sourceware.org
mailing list for the GDB project.
Re: [patch] Fix new FAIL `reject p 0x1.1' [fixup]
On Mon, Aug 23, 2010 at 11:54 AM, Jan Kratochvil
<jan.kratochvil@redhat.com> wrote:
> On Mon, 23 Aug 2010 17:35:07 +0200, Doug Evans wrote:
>> On Fri, Aug 20, 2010 at 4:58 AM, Joseph S. Myers <joseph@codesourcery.com> wrote:
>> > On Fri, 20 Aug 2010, Jan Kratochvil wrote:
>> >> On Fri, 20 Aug 2010 09:34:30 +0200, Jan Kratochvil wrote:
>> >> 0x1.1 is a perfectly valid hexadecimal floating point. ?The new testcase:
>> >
>> > It's not valid in C source code (a binary exponent is required), though it
>> > is valid as input to strtod (like INF, NAN, NAN(n-char-sequence_opt) etc.)
>> > - is the intention here that GDB deliberately accepts something beyond
>> > what would be valid in C source code?
>>
>> I don't know what gdb is intended to accept.
>
> So far I believe GDB is intended to be more relaxed than the C compiler.
In general sure, but general rules need to be re-evaluated for each
context in which they're applied.
In this context ... I'm not sure, but I don't have a strong opinion either. :-)
> (PR symtab/11846 -> is accepted interchangeable with .)
> (static symbols get resolved from not-current CUs)
> etc.
>
> While thanks for catching it I still believe now my testcase update is the
> appropriate fix - if glibc supports then extended input syntax let the GDB
> user benefit from it.
It's ok with me.