TL5 alpha3 (rep.2)

[Barbara Beeton]

from staszek:
    BB> Barbara
    DC> David
    TE> Thomas

    BB> we recommend \usepackage{psamsfonts} to get a controllable result,
    BB> but even that's not perfect.  someone has reported as a bug that
    BB> lines don't break the same with and without psamsfonts (i.e., with
    BB> the type 1 fonts vs. pk's).

    \usepackage{psamsfonts} ?? Where it can be found?
    Sebastion, we haven't such file on TL5 alpha.

oops!  that should have been either
    \documentclass[psamsfonts]{amsart} (or amsproc or amsbook)
or  \usepackage[psamsfonts]{amsfonts}

so the facilities *are* there already if amsfonts.sty is present.

    TE> Sure, it does. I have used dvips and pdftex and compared the output of
    TE> bitmap fonts vs. scaled type 1 (map file trick mentioned above). No
    TE> difference in size.
    TE> 
    TE> I don't really understand why it works, but it does. :-) Maybe, we
    TE> should ask Tomas or Thanh.

    Dvips uses also tmf files and is enought smart to do the scalling.
    Quite the same result can be obtained in a way:
    cmbsy7 CMBSY10 ".7 EntendFont" <cmbsy10.pfb
    but, e.g.,
    cmbsy6 CMBSY7 <cmbsy.pfb
    is just scaled with more accuracy ;-)

    DC> You can in principle make the scaled 10pt match the 9pt metrics
    DC> exactly by wrapping it in a vf file

    BB> but providing the vf file is, like, work!  not really likely to
    BB> happen from ams, i don't think.

    Yes, but only making `good' vf files edited `by hand' and precisely
    defining the positioning has some sense.

this is the only way we do make vf files.  a *lot* of handwork in
some cases!  (many years ago, we implemented such files to be able
to use autologic times fonts in math.  determining the proper
metrics and building the files took a whole person-month, spread
over several months elapsed.  a *lot* of handwork ...  we were
never able to distribute these metrics because of a contractual
agreement with autologic; on the other hand we *do* have a real
appreciation of what is involved.)

    The only `legal' and 100\% clean way is using \font\X zzzz atYpt
    in the TeX file (it needs checking some packages). It can cure the problem
    of line breaks, etc.

this is what the psamsfonts option does -- but for people who really
want the metrics of a "designed" 9pt font, using, e.g.,
    wncyr10 at 8pt
isn't going to be happy.  fortunately, all the cm fonts *do* have
all the requisite sizes, and glyphs from math and symbol fonts are
usually used only one at a time.

    Other substitutions (in map file or in vf) are somehow artifficial
    and always can give impredictable results (e.g. some fonts can have 
    different set of glyphs, look at cmr5 and cmr7 ;-)

    For the moment, the approximation provided by bsr-missing-interpolated.map
    is better than nothing, and, again, making more stuff like new set of 
    vf fonts would make live harder. It would be better making some changes
    in style files since making ps and pdf files are more and more frequent 
    and expected.

i've just discovered a probable cause of problems ...
the "bundled" type 1 amsfonts collection on the ams server has *all*
the tfm files in it, not just the ones for the fonts that have type 1
renditions.  i will take that up internally; i think it should be
changed.  (but i don't always win these discussions.)

    But it is another question...

indeed.
							-- bb