"Ronald S. Bultje" <rsbul...@gmail.com> writes:
>>> +#undef vsnprintf
>>> +int avpriv_vsnprintf(char *restrict s, size_t n, const char
>>> *restrict fmt, va_list ap)
>>> +{
>>> + int ret;
>>> +
>>> + if (n == 0 || n > INT_MAX)
>>> + return 0;
>>
>> I don't think zero is the proper value to return here. It is not
>> consistent with any existing snprintf variant.
>
> Is -ERANGE OK for n > INT_MAX? I'm not sure what the user intended if
> n == 0, so not sure what to return either.
POSIX says EOVERFLOW for n > INT_MAX. For n == 0, I would assume the
user was expecting the standard behaviour.
>> These replacements should be in separate files so that they can be
>> enabled independently. --
>
> Will split in snprintf.[ch] and strtod.[ch] (in compat/). What would I
> use to test for HAVE_BROKEN_STRTOD? For now, just enable it if
> snprintf is not there and refine later?
Testing for a broken strtod() directly isn't possible, so we'll have to
check for the microsoft runtime by some other means. That's why I was
asking for some symbol, anything, that only this library defines. It
doesn't have to be a dedicated version symbol, just something that
nobody else has.
--
Måns Rullgård
m...@mansr.com
_______________________________________________
libav-devel mailing list
libav-devel@libav.org
https://lists.libav.org/mailman/listinfo/libav-devel
