Re: Checksums (was re: 8r fonts)

[KNAPPEN@VKPMZD.kph.Uni-Mainz.DE]

If one can encode about six letters in the checksum, we should think very 
carefull about which information should be encoded there. IMHO, encoding 
the font encoding is a particularly bad idea, since the tfm file format 
already has a place for a string called CODINGSCHEME.

This string should be as precise as possible and can of course be examined 
by dvi drivers. It can -- with pk fonts at least -- also be compared to the 
special information stored there. METAFONT *always* put the codingscheme 
special into the gf file, not only with mode defined specials.

There is other information which one likes to check (e.g. Version numbers)
and which is otherwise not accessible through the tfm file.

--J"org Knappen.



[knappen@goofy] tftopl logo10
(FAMILY MFLOGO)
(FACE O 352)
(CODINGSCHEME AEFMNOPST ONLY)
(DESIGNSIZE R 10.0)
(COMMENT DESIGNSIZE IS IN POINTS)
(COMMENT OTHER SIZES ARE MULTIPLES OF DESIGNSIZE)
(CHECKSUM O 37045067476)
[...]

[knappen@goofy]  pktype logo10.300pk
This is PKtype, Version 2.3 (C version 6.1)
'METAFONT output 1996.12.11:1503'
Design size = 10485760
Checksum = -124489922
[...]
330:  Special: 'fontid=MFLOGO'
345:  Special: 'codingscheme=AEFMNOPST only'
374:  Special: 'fontfacebyte'
388:  Num special: 15335424
393:  Special: 'jobname=logo10'
409:  Special: 'mag=1'
416:  Special: 'mode=cx'
425:  Special: 'pixels_per_inch=300'
446:  Special: 'blacker=0'
457:  Special: 'fillin=0.2'
469:  Special: 'o_correction=0.6'
487:  Postamble
488 bytes read from packed file.