problems with cm fonts -- action requested by knuth

[bbeeton]

dear colleagues,
as a consequence of an incompatibility found by oren patashnik between
the version of ccr10 on a linux slackware distribution and the canonical
version at labrea, don knuth has added a notice to his web page
    http://www-cs-faculty.stanford.edu/~knuth/cm.html
if you have not already read this notice, i ask that you please do so.

don is very highly displeased; his good will has been taken advantage
of and his explicit requests have been ignored.  this message is the
result of his request to me to initiate damage control.

thirteen files have definitely been identified as different from the
canonical versions at labrea:
   bigdel.mf
   bigop.mf
   cmbase.mf
   cscspu.mf
   greeku.mf
   italms.mf
   punct.mf
   roman.mf
   romand.mf
   romanl.mf
   romanu.mf
   romlig.mf
   symbol.mf
there may be others as well.

some investigation into the origin of the changes and the dates of
various distributions that contain the changes have already been
discussed on the tex-archive list and comp.text.tex.  while no more
changes have been identified, i believe it would be prudent to make
a complete check of all cm* (the 75 fonts mentioned in volume e) and
cc* files on ctan to make sure that no more differences are present.
likewise, i would recommend a check of the two cd-roms that have been
created under the auspices of tex groups, namely the 4alltex and "tex
live" cd's.  i will be willing to keep the records and send out notices,
if someone else is willing to do the checking -- are there any volunteers?

there may be ctan directories in which "inviolate" files at present
coexist with changeable files; this seems to be a bad idea, as it
becomes more difficult to monitor changes.  i will be discussing with
the ctan maintainers what might be done to tighten the controls.
oren has suggested that a master list (or lists) of "inviolate" files
in the form of a READ.ME file in the relevant directories would also
be helpful; i agree and will ask for help creating this documentation.

here are some excerpts from don's message to oren and me.

    Changes to kerning do not affect the check sum in a TFM file. I suppose
    that decision was a mistake, although I vaguely recall that we did it
    because there are many equivalent ways to specify the same kerning info to
    TeX and no efficient known way to decide when two are equivalent.

    [...] changed the character shapes in monospaced fonts. That doesn't
    strictly violate what I say on the copyright page. But I certainly think it
    was nasty of him to change something for the Mathematica book that I don't
    want to see in The TeXbook or The Art of Computer Programming, where I use
    the typewriter fonts a lot; he has changed the shape of the apostrophe and
    reverse apostrophe! He should have given his fonts a different name, and
    not foisted them off on people like me who want the normal cmtt10.
    ([The] changes of i and I for "cmsstt" fonts are no problem, but I think
    such fonts would need even more changes before they become acceptable
    aesthetically.)

    After thinking about it some more, I believe it should be considered
    illegal for anybody to change the official CM sources, just as it's wrong
    to change the TeX or MF sources that I am responsible for. The copyright
    page on Volume E explicitly allows people to tune up the fonts in any way
    they like provided that they leave the TFM files alone, but I was thinking
    of tuning bitmaps or making outline fonts, or making private versions of
    the MF sources that would not override the official files described in
    Volume E. Those files should always be the standard distribution.

    [...]  All we need is for somebody to figure out exactly which files
    are not verbatim copies of the sources at labrea. Those files are in error
    and must be suppressed.

    [...] We need to [...] get the word out to everybody who has received these
    bad font sources, so that they REPLACE them pronto... I will put something
    on my web pages, and I'll update it as more information is available.
    We need to find everybody who distributes those bad font sources and get
    them correct their files immediately.

another concern don had (although i'm pretty sure it will turn out
okay) is the use of the sauter parameter files rather than the separate
driver files, cmr10.mf, etc.  don has requested that the 75 tfm files
corresponding to the fonts in volume e be checked against the correct
ones (the ones at labrea),
    (By doing TFtoPL and then a file compare.)

    And I still believe nobody should diddle with the files accent.mf,
    greekl.mf, etc. that I maintain...

please get in touch if you are willing to take on part of this task.

don has also been approached by someone who feels that this might be
a fit subject for a news story:

    [...]  she thinks the San Jose Mercury News might want
    to have a news story about this. I told her that I wanted to hold off until
    the TUG board has reached some sort of consensus about what should be done.
    If the board thinks that newspaper publicity will help spread the word
    about the problem and thereby lead to a faster purge of bad data, I'll tell
    her to go ahead and promote such a story.

i'd appreciate comments on this as well.

i'm not including here the details of what's been changed, but will
provide that and other information to volunteers as needed.

checklist of tasks:
 - compare cm* .mf files, labrea vs. ctan (75 volume e fonts)
 - compare cc* .mf files, labrea vs. ctan
 - identify cm* and cc* .mf files at ctan outside the "knuth" area
 - check .tfm files for cm* fonts, labrea vs. sauter
 - check .tfm files for cc* fonts, labrea vs. sauter
 - repeat all checks for 4alltex cd
 - repeat all checks for "tex live" cd
 - identify directories that contain both "inviolate" and "variable" files
 - create READ.ME files for "knuth" areas listing "inviolate" files and
   stating the reasoning behind the restrictions
 - determine whether any rearranging of the ctan (or tds) areas are
   necessary or desirable
 - determine what other collections (e.g. linux slackware distribution)
   need to be checked

the last item is somewhat open-ended -- i'm not familiar with all the
distributions that may exist -- so please, if anyone knows of such a
collection, whether it's posted on the net somewhere or distributed on
a cd, please send in the information.

i hope this can be well under way, if not complete, by the time i leave
on july 23 for the tug meeting.
						-- bb