Behaviour of \latinfamily

[Rebecca and Rowland]

At 11:45 am +0200 28/5/98, Thierry Bouche wrote:
>Concernant « Re: Behaviour of \latinfamily », Rebecca and Rowland écrit : «
>» Any idea what this ETXFILE.etx is?
>
>an encoding file described in fontinst's obscure slang.

Well, yes; I was wondering exactly how fontinst knows that an etx file
needs to be read in and how it decides which particular etx file to read.
According to Ulrik, the comment in the file is mis-placed and \afmtomtx
doesn't read an etx file at all; \pltomtx does, and it selects the etx file
based on the various \declareencoding commands.

>» But this can't be right: fontinst doesn't create tfms at all.
>
>when saying tfm, i meant of course PL, installfont makes a VPL whereas
>installrawfont makes a PL.

Righto - ta.  One thing about \installrawfont.  I gather that 8r encoded fd
files are not supposed to be created (dunno why: I'd find them useful), and
\intallrawfont has this syntax:

%    \installrawfont{FONT-NAME}{FILE-LIST}{ETX}
%       {ENCODING}{FAMILY}{SERIES}{SHAPE}{SIZE}

as I understand it, the last five arguments are only used in writing to an
fd file.  If 8r encoded fd files aren't made, what's going on?  I've
obviously mis-understood something here.

>» This doesn't tie up with my observations: I've never seen psyr specified
>» anywhere, and this doesn't seem to take into account the various fakeries
>» fontinst can do, or take into account any (say) founts with small caps that
>» might be present which Sebastian tells me will be used if they exist.
>
>I meant _in your simple case !_ \latinfamily has nothing to do with
>psyr. oblique (resp. small caps) fonts are faked by
>\transformont{\slantfont (resp. simply using another encoding file
>T1c.etx say, that declares that Asmall is A scaled down), other
>fakeries, as Scaron or the FFsmall lig (!!!!) are made by latin.mtx.

Right, but I was asking about \latinfamily.  I'm wondering which founts
\latinfamily looks for when it's trying to find a small caps fount (amongst
other things).

>Your psyr makes me think that you're talking of mathptm, in which case
>you simply don't usde \latinfamily, but lower level macros like
>\installfont, where you only have to give various metrics (mtx) that
>can declare characters, kerning, ligatures, and fake almost anything;
>an encodings (etx) that tell fontinst where to place things.

Righto.  But I didn't mention psyr in the first place; you're the one who
said that you thought \latinfamily was equivalent to a load of stuff which
included psyr.  I've never used psyr.mtx and I've never used mathptm. But
never mind: you've given me some more useful information which has been
stashed in my `file of useful notes on fontinst'.

[snip]

>» To my mind, installing a fount for use with LaTeX involves creating tfm and
>» vf files, moving them to a place on the TeX and dvi driver search paths
>» respectively, creating useful fd files, and adding lines to the dvi
>» driver's fount mapping file.  Of this, fontinst does one job only when you
>» execute a \latinfamily command: creating useful (sometimes) fd files for T1
>» and OT1 encoding, *not* 8r encoding.  So fontinst doesn't ever install
>» founts.  This is why I ask what you mean when you say `install ... fonts'.
>
>well, writing an FD file is in some sense registering that new font
>set in latex nfss. producing PL/VPL files is making this font set
>available to tex (after pltotf, vptovf...). It's the non-trivial part
>of the installation, all the rest may be done by perl, shell scripts,
>hand -- whatever. As i don't speak english very well, i can't garantee
>what i write here at that level of linguistic accuracy.

I greatly appreciate the time you've spent answering my posts, but I think
this is a cop-out: your written English is *excellent*.  I think the
problem is that all you fontinst experts *think* in short-cuts which are
blindingly obvious to you, but not to someone like me.  Not that it
matters: I'll get where I'm going eventually.  Probably.  I've just had a
look at the `fontinst documentation so far' file: 42K and half of it's
rubbish.  And it's yet another wet morning in Lancashire.

(I think there was some point to that: the thing is that I know what
fontinst does in broad terms, but to produce useful documentation, I need
to know *exactly* what it does *in detail*.  I'm beginning to think I might
be close to being able to work it out for myself.)

Rowland.