pdftex bug

[Vincent Lefevre]

On 2023-09-13 22:35:41 +0900, Norbert Preining wrote:
> On Wed, 13 Sep 2023, Ulrike Fischer wrote:
> > Well I wouldn't count it a pdftex bug if adobe reader can't handle 
> [...]
> > here. Imho they simply copy the values from the .pfb. You could try
> 
> Not that I know, but maybe the specification of type1 fonts requires
> a certain maximum precision.

Here, 0.000833333333333 and 0.000833333333 are just decimal strings.
Aren't the computations done in binary internally (with native
FP types)? If something like strtod() is used, the *decimal*
precision shouldn't matter.

What might happen is that the bug appears with some value, but not
with another one, but this is just a question of luck.

Note that 1 / 0.000833333333333 ≈ 1200, so that it is possible that
the code happens to fail if you get too close to 1 / 1200, possibly
due to a combination of a rounding error and the use of floor/ceil
function. FYI, GNU Emacs had such a bug in the computation of the
font metrics: https://debbugs.gnu.org/cgi/bugreport.cgi?bug=44284

-- 
Vincent Lefèvre <vincent at vinc17.net> - Web: <https://www.vinc17.net/>
100% accessible validated (X)HTML - Blog: <https://www.vinc17.net/blog/>
Work: CR INRIA - computer arithmetic / AriC project (LIP, ENS-Lyon)