laser-lovers@uw-beaver (10/08/85)
From: "Francie J Newbery" <fjn@Purdue.EDU>
I have a question about the post-processors for ditroff. We have
one that works fine for us with one exception. I was hoping that
maybe someone already knows about this bug and can suggest a fix.
An example of the problem can be seen when we run a file which contains
many super or sub-scripts and has equation numbers through Eqn, Troff -ms and
dlgp (symbolics post-processor). The equation numbers shift further
and further to the left the more super or sub-scripts you have.
Here's a short example of a file that messes up.
.EQ 1
x = f(y/2) + y/2
.EN
.EQ 1
x sub i x sub i x sub i x sub i x sub i x sub i x sub i x sub i
.EN
It seems like maybe each super or sub-script causes a small truncation
error somewhere, and after many of them the equation number shifts noticeably.
Perhaps some strange business having to do with scaling?
I am not sure whether the problem is with ditroff or with the postprocessor.
We also have some postscript printers and the problem *does* appear there
as well (using the Transcript psdit filter).
With both the symbolics and the postscript printers, the output appears
exactly correct when using the old C/A/T troff and it's filters.
Francie Newbery
Facilities Staff, Purdue University CS dept. (fjn@purdue.EDU)
{inhp4, allegra, decvax}!pur-ee!purdue!fjnFURUTA@WASHINGTON.ARPA (10/27/85)
From: ihnp4!dual!sunybcs!colonel@uw-beaver.arpa (Col. G. L. Sicherman) There are still versions of eqn around that don't check for -Tdevice (and users who don't specify it). This is a pretty reliable way to get truncation errors....