[9front] troff & eqn sqrt{} output as pdf

[9front] troff & eqn sqrt{} output as pdf

[Ethan Long]

Hello, unfortunately when using equations in troff there is one
bug/erratic behaviour that I cannot fix, so I was wondering if anyone
here would be able to help.

Basically when using the sqrt function in eqn, everything appears to
work okay and well if you're outputting as ps and viewing in page on
9front.
However, when one uses ps2pdf or views this postscript elsewhere
(ps2pdf on 9front and linux both do this), the horizontal line that
typically covers the whole square root is detached from the radical
that starts the root.

Try this example:

eqn file | troff -ms | dpost | page
eqn file | troff -ms | dpost | ps2pdf | page

.EQ
sqrt {g} = pi
.EN

It doesn't matter how long the root is, you will notice that if you
simply run dpost and pipe into page all is fine, but if you use
ps2pdf, the top of the root is offset by a large gap.

Does anyone have any ways of fixing this or working around this? If
it's a ghostscript bug then it's been there a long while.

Re: [9front] troff & eqn sqrt{} output as pdf

[Steve Simon]

there is a formatting bug in tbl - which i think has been fixed in the heirloom code. i don’t know if this is related but small vertical shifts so horizontals and verticals don't meet is very reminiscent of that problem.

-Steve

Re: [9front] troff & eqn sqrt{} output as pdf

[umbraticus]

So we discovered that this change:

diff -r 6086ad8502ee sys/lib/postscript/prologues/dpost.ps
--- a/sys/lib/postscript/prologues/dpost.ps	Fri Jan 15 19:32:02 2021 +0100
+++ b/sys/lib/postscript/prologues/dpost.ps	Sun Jan 17 22:17:45 2021 +1300
@@ -188,1 +188,1 @@
-	/radicalex		[-50 0 500 0]
+	/radicalex		[-50 0 500 1]

makes this:

; echo '.EQ
sqrt {g} = pi
.EN' | eqn | troff -ms | dpost | ps2pdf | page

work but have no idea why.
Making the changes suggested at:

B /sys/lib/postscript/prologues/dpost.ps:169,176

didn't help but changing that 0 to 1 (or anything else!)
seems to do the trick for whatever reason.
Any thoughts? or just push it?

Ethan → can confirm on a 5 page report that it fixes all instances of roots

I also checked man -t eqn | dpost | ps2pdf | page

umbraticus