This is the mail archive of the
libstdc++@gcc.gnu.org
mailing list for the libstdc++ project.
Re: valarray_copy implementation question
- From: Gabriel Dos Reis <gdr at integrable-solutions dot net>
- To: Paolo Carlini <paolo dot carlini at oracle dot com>
- Cc: Christian BRUEL <christian dot bruel at st dot com>, libstdc++ at gcc dot gnu dot org
- Date: Mon, 16 Nov 2009 16:04:08 -0600
- Subject: Re: valarray_copy implementation question
- References: <4B016F1E.1030505@st.com> <4B01C805.20600@oracle.com>
On Mon, Nov 16, 2009 at 3:45 PM, Paolo Carlini <paolo.carlini@oracle.com> wrote:
> Hi,
>> Hello,
>>
>> The standard interpretation that lead to the implementation of
>> __valarray_copy used for the mask_array assignment in the valarray
>> looks strange.
> ...
>
> A few comments. First, certainly it would help to have a small but
> self-contained testcase, building but either segfaulting or triggering
> valgrind errors. Comparing / constrastring to other implementations
> would also help. Moreover, I'm seeing the pattern using 'while (!
> *__ok)' in tons of other functions, are those related or not?
>
> That said, I would like to solicit feedback from Gaby Dos Reis, the
> original author of this part of the library...
I did not consider the interpretation put forward by Christian, although I can
see his point. I suspect I have to double-check what Fortran
does in this case and consider whether it actually is a LWG issue.
Do you know what other implementations do?
-- Gaby