Re: AFMs for Blue Sky CM/PS fonts...

["Berthold K.P. Horn" <bkph@ai.mit.edu>]

   >It is hard to imagine how that could be unless your PL files are
   >also corrupted, since these AFM files have the wrong advance widths.

   What do you mean by saying `advance width'? A glyph has afaik two different 
   `widths':

   a) The tfm width, which is measured in TeX scaled points (typically) and
      is resolution independent.

   b) The value chardx, which is the real advance taken by the renderer 
      (printer or screen) and which is necessarily an integer number of 
      pixels.

   The latter value is not relevant for TeX tfm files, but is recorded in the 
   gf and pk files. If the you mean the latter value, it can be corrected in 
   the afm files without changing the related tfm files.

I am not concerned about bitmaps (as you know :=), and beyond that
there is only one `width.'

This should be identical in the AFM file and the TFM file.  And since
the widths in CM fonts are not integer multiples of 1/1000th of an
em, they must be non-integers in the AFM file.  Any AFM file for CM
fonts that has integer widths (except for a few special character at a
few special sizes) is wrong (not by much, but by enough that it matters).

Any program purporting to make TFM files from AFM files should then
also get the wrong width in the TFM files if it working from these 
rounded widths.  

There is another issue, which is that the widths in the TFM files
for CM themselves are not accurate, relative to Knuth's design.
The bottom 2 or 3  bits are wrong, probably due to arithmetic round-
off in METAFONT, but this error is very small compared to the above.
One can tell easily, since for e.g. for some fonts most dimensions are
supposed to be integer multiples of 1/36pt (or sometimes 0.1 * 1/36pt).

Regards, Berthold.