[tex-live] Charter fonts have different, but identically named AFMs.

[Karl Berry]

    a third set of AFMs in tlpkg/tlgs/fonts/,
    with matching PFB files in that same directory. Those PFBs are
    identical to the ones in type1/bitstrea/charter/

Ack.

    One assumes that the mathdesign AFMs were used as source files
    in the generation of the mathdesign TFMs and VFs, so perhaps
    they shouldn't be changed, 

Ack.  But we should probably tell Pichareau, eventually.

     the AFMs in
       afm/bitstrea/charter
    should match the PFBs in
       type1/bitstrea/charter/.

Ack.  I copied tlgs/fonts/bch*.afm to
texmf-dist/fonts/afm/bitstrea/charter/ and committed them.

    Possibly also the TFM and VF files in the charter package would need
    to be recreated afterwards.

Now this I do not want to do with testing.  The existing mathdesign
fonts, based on these "correct" afms, way overkerns punctuation, e.g.,
the pair "V." (value about -222).

The tfm's we have in fonts/tfm/bitstrea/charter have a good kerning
table (value about -78 for V.).  I'm sure they came from a release of
the Charter fonts by Bitstream (probably the original release) and then
later Bitstream released different afm's.

I discovered this when working on the interview book
(tug.org/interviews), which uses the mathdesign fonts, and the kerning
was just too horrible to print with.  I ended up writing a little script
to copy over the kerning table from the good tfm into the mathdesign tfm.
(tug.org/interviews/book/kernfix)

Maybe the underlying problem is the mismatch with the PFB's and that if
we generated the TFM's from the AFM's and used the right PFB's, the
kerning would be ok.  But it needs to be checked.  Is that something you
could try out?  (Just thought I'd ask!)

Thanks,
Karl