possibly severe bug in fontinst 1.914

[Lars Hellström]

At 15.48 +0100 2001-03-06, Walter Schmidt wrote:
>Hi,
>
>the following sequence of commands works fine with
>fontinst 1.801:
>
>\transformfont{padr8r}{\reencodefont{8r}{\fromafm{padr8a}}}
>\fromafm{padri8x}
>\installrawfont{padr8r}{padr8r,8r}{8r}{8r}{padj}{m}{n}{}
>\installfont{padr9o}{padr8r,padr8x,latin}{OT1j}{OT1}{padj}{m}{n}{}
>
>However, when I use fontinst 1.914, a file padr8r.pl
>with a positive slant value is written:
>
>(FONTDIMEN
>   (PARAMETER D 1 R 0.334)
>
>Apparently, the slant valus from the preceding "\fromafm"
>command finds its way in.  Do I use \fromafm in a wrong way
>here, or is this a bug?

I'd say you use it in a wrong way. \fromafm is a wrapper around \afmtomtx
which additionally calls \mtxtopl and sets the afm-name string (for
\mtxtomtx). As it is a <transformed font> command, it should only be used
in the second argument of \transformfont. Actually I don't think you need
to use it explicitly; \installfont will generate the MTX if it cannot find
one (there could be some issue if the generated PL still hangs along,
though).

OTOH there really is a bug, since the same thing would happen if you used
\afmtomtx directly. This is a consequence of that \setint and friends now
do global assignments between \installfont and \endinstallfont (as long as
you're not inside a \transformfont, \installfont, \installrawfont, or
\reglyphfonts ... \endreglyphfonts); apparently I'll have to reset this to
local assignments in a few more places than I thought.

>PS:  fontinst is a GREAT piece of software; so I hardly
>dare to complain about the few remaining bugs...

Well, thank you.

Lars Hellström