[tex-live] dvips updates in source.development

[Karl Berry]

Since not everyone here is on tex-k, I thought I'd mention that Tom has
checked in some noteworthy fixes to dvips.  (You can see the full msgs
in the tex-k archives if you want them.)

k


From: "Tomas G. Rokicki" <rokicki at CS.Stanford.EDU>
Date: Sat, 15 Jan 2005 18:36:48 -0800

I just submitted a fix for the problem where a document includes a
figure that uses a downloaded PostScript font, and that same font
is also used by the main document itself.  Dvips was partially
downloading the font in this case when it should have downloaded
the entire font.  It now does download the entire font.

[...]

I've got a few more changes to make; I'll try to get them done as
soon as possible.

But I'd be slightly hesitant to package the *current* dvips in
the p4 tree into any distribution for widespread release until we
get some testing in.  If you need a dvips for your distribution,
grab the one from yesterday instead.


Date: Sat, 15 Jan 2005 16:28:36 -0800
From: "Tomas G. Rokicki" <rokicki at CS.Stanford.EDU>
To: tex-k at mail.tug.org

Howdy, everyone!

I recently received some mail from a mathemtician using epic/eepic to
render some very sweet graphics.  Unfortunately, dvips was rendering
these graphics very badly, mispositioning some elements significantly
(about a quarter of an inch in some cases).

The problem turns out to be the pervasive use of rounding and integers
throughout drawPS.c.  This code is 20 years old, so the bug has been
there all along.  Especially with spline paths, error can accumulate
with all the integer roundings causing severe mispositioning.

To fix this, I have a new version of drawPS.c that eschews integer
math throughout, using doubles almost exclusively.  On the diagrams
I've been able to find, it looks fine.  But the changes are somewhat
extensive if mechanical, and I'm terribly afraid to break other graphs
because of a typo or some other mistake on my part.

Is there a repository of eepic/epic graphics that I can use to gain
more confidence in the fix?  Perhaps someone familiar with this macro
package would be willing to take a look at my changes?

[...]