On Thu, 14 Dec 2017, Jakub Jelinek wrote:
> On Wed, Dec 13, 2017 at 03:55:13PM -0700, Martin Sebor wrote:
> > For the second part, can you please also add a compile-time test
> > to verify that the result isn't constrained to the same range as
> > with a real argument? Checking that the abort bel
On Wed, 13 Dec 2017, Jakub Jelinek wrote:
> Hi!
>
> This patch fixes 2 issues in format_floating. One is that when determining
> precision, we should consider solely the type *printf* will read the
> argument as (i.e. double unless L or ll modifier is used, in which case
> long double), not the
On Wed, Dec 13, 2017 at 03:55:13PM -0700, Martin Sebor wrote:
> For the second part, can you please also add a compile-time test
> to verify that the result isn't constrained to the same range as
> with a real argument? Checking that the abort below isn't
> eliminated would do it for %f:
>
> vo
On 12/13/2017 03:16 PM, Jakub Jelinek wrote:
Hi!
This patch fixes 2 issues in format_floating. One is that when determining
precision, we should consider solely the type *printf* will read the
argument as (i.e. double unless L or ll modifier is used, in which case
long double), not the type of
Hi!
This patch fixes 2 issues in format_floating. One is that when determining
precision, we should consider solely the type *printf* will read the
argument as (i.e. double unless L or ll modifier is used, in which case
long double), not the type of the argument, because the corresponding
argumen
