Behaviour of \latinfamily

[Rebecca and Rowland]

At 11:03 am +0200 2/6/98, Ulrik Vieth wrote:
>>> It tries all the weights in \latin_weights, all the shapes in
>>>\latin_shapes,
>>> and all the widths in \latin_widths.  (Well, not really.  There's a switch
>>> that determines whether fontinst attempts to fake narrow fonts, or whether
>>> it only uses those narrow fonts, which are provided as real fonts.)
>
>> Can you explain what the switch is, and what you need to do to change it?
>> Does fontinst ever change this switch on its own?
>
>IIRC, this switch is called \if_fake_narrow.  It is never changed by fontinst
>itself, but there is a macro \fakenarrow{WIDTH FACTOR} to change it.

It certainly looks like it:

\newif\if_fake_narrow_
\_fake_narrow_false
\def\fakenarrow#1{
   \_fake_narrow_true
   \gdef\fake_narrow_width{#1}
}

As far as I can tell from reading the source code, this controls whether or
not \latinfamily will attempt to fake a narrow fount; if \if_fake_narrow_
is false, it'll happily create vpls for `real' narrow founts if you've got
them under any circumstances.  Is this right?

>>> (BTW, the present CTAN version was missing some weights, which were
>>> simply ignored.  In addition there were some weight mapping which lead
>>> to problems, such as regular (or book) and medium being mapped to the
>>> same LaTeX weight.  I don't know when Sebastian will put up the new
>>> version, in which these problems have been fixed.)
>
>> Is this mapping the `m' weight (fontinst name) to `mb' (fd file weight)?
>
>Yes.  The `mb' designation was my idea, since it was inappropriate to
>map both 'r' nad 'm' to the same code.  Presumably, it was mistake that
>NFSS codes used 'm' for regular in the first place, but it's too late
>to change it now.

Good stuff - thanks for this.

Rowland.