On Sun, Oct 04 2020, Bruno Haible <bruno@clisp.org> wrote:
> Hi Jeremie,
>
>> The attached patch changes vasnprintf.c to avoid using %n in the general
>> case, ie when the return value of snprintf is usable.  This should help
>> if more systems decide to make tighten %n usage.  There are plans for
>> that in OpenBSD land.
>
> Thanks for the suggestion. Yes, this is a good idea. (I didn't know about
> the OpenBSD plans.)
>
>> The existing comments in vasnprintf.c mention systems where
>> gl_SNPRINTF_RETVAL_C99 and gl_SNPRINTF_TRUNCATION_C99 pass.  This patch
>> only considers the usability of the return value of snprintf, as lack
>> of truncation seems to be a different problem (apparently handled later
>> in the code).
>
> I prefer to limit the shortcut to those platforms where both
> gl_SNPRINTF_RETVAL_C99 and gl_SNPRINTF_TRUNCATION_C99 pass. These are
> the conditions I found when considering this code in detail a couple of
> years ago. Maybe you are right that only one of the conditions is needed;
> but I don't want to live on the risky edge here.

Your approach certainly makes sense.  I considered something similar
earlier.  Your changes look good to me, thanks!

-- 
jca | PGP : 0x1524E7EE / 5135 92C1 AD36 5293 2BDF  DDCC 0DFA 74AE 1524 E7EE