Re: [Groff] Bug in devdvi - found!

[Werner LEMBERG]

> Found the problem (I think).  Line 14 of dvi.tmac is missing an
> entry for EX.  It should read something like:
> 
> .special TR TI MI S EX SA SB CW
> 
> (I'm not sure which of the symbol fonts should be searched first).
> When this is added, all behaves normally, \(is gives the proper
> integral sign, and (for eqn) int from 0 to inf has everything in the
> correct fonts.

With the eqnrc patch given in another mail, this is no longer
necessary.

> I'm unclear why CW should be declared here as a special font;
> wouldn't it be better to do .fspecial CW CW ?

>From fonts/devdvi/generate/Makefile:

  # TR is special because it contains \(pl \(eq
  # TI is special because it contains \(Po
  # CW is special because it contains "

> Some of the .fspecial requests immediately following look a little
> odd to me.  Why should TI (= Computer Modern Italic) have CWI (=
> Computer Modern Typewriter Italic) as a special font?

To get `"' from CWI if TI is active.

> TBI also has CWI as a special font, which doesn't make sense to me
> either.

There is no Computer Modern Typewriter Bold Italic.

> Should the first of these lines (i.e., line 15) in fact read:
> 
> .fspecial CWI CWI
> 
> so that the special glyphs at the beginning of the CWI font can be
> found?

Mounting CWI as a special font while CWI is active is completely
meaningless.  groff always searches glyphs in the current font first
(except for entities defined with the `char' request).

> Finally, the file
> 
> .fam C
> .LP
> abcde
> 
> produces the error (with the dvi device):
> 
> try.ms:1: warning: can't find font `CR'

devdvi contains only two families, `T' and `R', no `C'.  cmtt itself
is already quite bold; while bold cmtt exists (but neither in the `cm'
nor in the `cmextra' bundle on CTAN, thus not in most teTeX or TeXLive
distributions), bold italic doesn't AFAIK.  It is very ugly, BTW.

> yet dvi.tmac contains the line .ftr CR CW.  Why should this be?

This has historical roots only.  Compare it with the entries in
`ps.tmac'.


    Werner