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

[Vladimir Volovich]

"AJ" == Alan Jeffrey writes:

 >> And while we are at this topic, some more bug. :-) This time it
 >> can be called something like "rounding problems?".

 AJ> Hmm yes, this is another hard-to-fix problem.  Looking to see
 AJ> whether a font has already been loaded is easy (just do a csname
 AJ> lookup on <font>-<fontsize>) but doing an approximate match is
 AJ> much harder!

Moreover, why these different numbers could ever occur? They should
not, because the calculations, though approximate, always give the
same result. Something is wrong here. Maybe, somewhere differing
constants for font sizes are builtin into fontinst?

 AJ> For most fonts it won't make that much difference, the only
 AJ> problem is the extra mapfont line and perhaps some inefficiency
 AJ> since font 0 is the default font.

 AJ> Are the ae fonts doing anything weird to generate that extra
 AJ> MAPFONT line?

Not at all, from my unexperienced point of view. :-) Here is how one
can reproduce this problem with all the standard fontinst 1.801 files
(nothing ae-specific):

1) create a file "test.tex" which contains the following lines:

---------------
\input fontinst.sty

\declareencoding{TEX TEXT ITALIC}{ot1i}

\installfonts
  \installfamily{T1}{aer}{}
  \installfont{aer17}{cmr17,kernoff,cmmi12,cmu10,cmsy10,latin}{T1}
     {T1}{aer}{m}{n}{<17->}
\endinstallfonts

\bye
---------------

2) create a file "go" which contains the following lines:

---------------
#!/bin/sh
tftopl cmmi12 > cmmi12.pl
tftopl cmr17 > cmr17.pl
tftopl cmsy10 > cmsy10.pl
tftopl cmu10 | sed 's/TEX TEXT/TEX TEXT ITALIC/' > cmu10.pl
tex test
---------------

3) run the shell script:

sh go

4) look into the generated aer17.vpl

	Best regards, -- Vladimir.