bug in fontinst 1.801: duplicated kernings (was: ae fonts Q)

[Rebecca and Rowland]

At 5:03 pm +0100 28/9/98, Alan Jeffrey wrote:
>>  LE> I can't remember why, and when I removed the % signs, the fonts
>>  LE> still seem to be generated OK, so I would call the lack of
>>  LE> kerning a bug.
>>
>> Hmm... i've just checked, and found that when those lines are
>> uncommented, fontinst generated duplicated kernings, e.g.:
>
>Well, you can call it a bug, if you add kerns between a pair of
>already kerned characters, probably fontinst should ignore the 2nd
>kern pair.  But that would add a run-time penalty to an already
>painfully slow program.

I don't see this as a big problem myself: fontinst is slow for sure, but it
runs at a manageable speed on my old computer (a Mac Performa 475 with
roughly the processing power of a 40MH 486SX).  A slight slow-down is
neither here nor there, especially when you consider that most computers
these days are an order of magnitude faster than mine, and computers are
getting faster by the day.

>Best bet would be to post-process the fonts
>in perl.  This is already a good idea for some other fonts which
>generate huge easily-compressed kern tables.

This is fine for people who can use a Perl script to do this job.  The
problem is that Perl isn't very portable: the vfinst Perl scripts, for
example, don't work on my Mac although I do have a Perl port (the scripts
assume, for example, that my vptovf (etc) can accept command line arguments
which they don't on Macs.  Short of learning all about Perl and re-writing
the scripts from the ground up, they're unusable on Macs).  So while this
is okay for particular individuals, it's a bad idea for general consumption.

Rowland.