[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: comments on mathfont-0.5



Matthias Clasen:

> I have already changed the format of the CHANGES file (yesterday, just
> after putting 0.5 on the net). Perhaps I should find out how to convince
> emacs and cvs to create a nicely formatted ChangeLog...
> Is a date format like `May 14, 97' acceptable ? Btw isn't 97-05-14 the
> official (ISO) format for that date ?

Typing "C-x 4 a" opens or creates a ChangeLog file.  I don't know how
this interacts with CVS but there should be some way if you RTFM...

> Well, I might have messed up that file. When I first wrote it, I noticed
> that the font tables for the extensible fonts went *way* of the page,
> so I hacked nfssfont.tex to split the table in two halfes. Maybe that 
> confuses other parts of nfssfont.tex

Ideally, nfssfont.tex should split the tables across pages automatically 
just like Knuth's testfont.tex does.  It all depends on a reasonable 
setting of \textheight.  Using 650pt = 9in should be fine for both letter 
paper and a4paper.

> This is really a bug. I probably have always looked at the mf sources
> in the various working directories. The easiest solution would probably
> be to simply omit the copying and leave the mf sources in the various
> subdirectories. On a second thought, the files in mathfont/mf are
> really useless, they will never be used by the make process unless you
> remove their originals from ym[a-f].

Well, the problem is MakeTeXPK triggered from dvips or xdvi.  After
sourcing PATHS form the shell, all subdirectories are searched and the
first file with matching name found will be used.  Duplicate filenames
are always a problem and should absolutely be avoided, as it is cleary 
pointed out in the TDS standard.

> The main reason for arranging the MX fonts like they are now is 
> to have all slots which are directly accessed by macros at the beginning. 
> This is also the reason for the strange separation of the pieces of
> some extensible recipes. Frankly, I do not think that a nicely arranged
> font table is such an important factor. 

Well, it cetainly helps to see how many sizes of a certain delimiter
are availabe.  Sure, it's just a cosmetic issue, but it may make the
difference as to whether a font table will be perceived as messy or
organized when it will be eventually presented to the general public.
(And it should be presented, perhaps at EuroTeX'98 in St. Malo???)

> Since you have proposed a more traditional layout of MX, I'd like to 
> hear your opinion on the xm[bf]s fonts which offer only the traditional
> range of larger sizes. I decided to keep the splitting in MX1/MX2.
> This has the big advantage that the macros can be the same - just the
> variable parts of the font encodings are different and these are not
> directly accessed by macros.

Well, one reason for having a more traditional layout of MX was to
keep the number of math families (or \mathgroup's in LaTeX) restricted
to 4 or 6 for compatiblity with plain or AMS TeX.  Besides there is
only one \fam3 slot where the required \fontdimes are taken from.
Thus having a reduced MX1/MX2 combo won't help.  I'd still vote for
a single (reduced) MX in the basic set and and optional (extended) 
MX1/MX2 combo to replace it in an optional add-on package.
 
> One more remark/question concerning bold: How do you like the bold
> versions of the MX fonts ? I have created the parameter files by
> diffing cmsy and cmbsy and applying similar changes to the cmex
> parameters. Is it a good idea at all to have bold extensibles ? On the
> other hand, if we don't, the basic size delimiters from MC will be bold
> while their larger cousins are not, same for accents.

If I recall correctly, cmex has almost the same parameters as cmr.
Likewise cmmi and cmsay are mostly the same. (That is, apart from
fontdimens and the font identifier).  See my comments on hacking
conrete math in CTAN:macros/latex/contrib/other/concmath/HISTORY.

I haven't yet seen your bold MX fonts them as the bold MX tables
somehow weren't included in charts.{tex,dvi}.  The boldness should
probably only affect some shapy delimiters while some other geometric
delimiters should be unchanged.  Just look at Knuth's euex to see
which characters are modified and which are not.

Personally, I hardly ever care about a \mathversion{bold} where the
boldness applies to everything, except for the few (usually fairly
simple) bits of math that might sometimes appear in section headings.
What I really need is a bold math alphabet for vectors, where the
boldness applies only to letters and letter-like symbols but leaves
\mathbin's, \mathrel's, bigops and delimiters unchanged.

In a conventional CM set-up, that would mean something like:

\DeclareSymbolFont{vectors}{OML}{cmm}{b}{it}
\DeclareSymbolFont{tensors}{OT1}{cmss}{sbc}{it}
\DeclareSymbolFontAlphabet{\mathvec} {vectors}
\DeclareSymbolFontAlphabet{\mathtens}{tensors}

> Sadly, there is not enough metaness in the AMS symbol fonts to create
> bold versions easily. I have extracted the few glyphs which produced
> acceptable bold versions and added them to yma (kappa1, backepsilon, 
> hbar, hslash,...)

As I said, I personally don't care much about making erverything bold.
In a lecture manuscript of 500+ pages of quantum mechanics I never
ever needed such a weird thing as a bold \hbar.  Shudder!

Cheers, Ulrik.