[tex-live] Small scaling values in xetex / (x)dvipdfm(x)

[David Carlisle]

On 9 May 2017 at 06:06, 平田俊作 <shunsaku.hirata74 at gmail.com> wrote:

>
> Hi,
>
> I think Acrobat disliked such transformations,
> giving an error for such cases.
>
> Shunsaku Hirata
>
>
> I tried with acrobat reader DC and got reasonable results even with a
scale factor of 0 (inserted via pdf:literal)

With a (now very old) copy of acrobat 9 pro I did get some strange effects
if the scale factor was very small
but I don't think that  if the viewer is doing arithmetic using double then
1e-5 is particularity small and we shouldn't
make strange output on all systems just for old versions of acrobat.

It (might:-) be interesting to know how old a version of acrobat you have
to have before this is a problem.
Note that for the last 18 months latex has avoided this dvipdfm issue by
applying the scale factor via a pdfliteral
cm matrix operator rather than the driver's x:scale special, and no one has
reported any issues with acrobat.
(Although I admit probably not many people are scaling things by these
small values)


If we wanted to avoid small scale factors (personally I'd do this also for
0) I'd give the warning and force it to be
the smallest "safe" value, not just simply not scale, as currently.


But for values within the normal range of tex dimen <factor> (that is, 1sp
= 2e-5pt) I don't think there should be any
warning or adjusting of the values at all. Akira's change just now would do
that,

David
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://tug.org/pipermail/tex-live/attachments/20170509/b1d4780d/attachment.html>