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

Re: V0.4 details

Joerg Knappen:

> Here are some details to the 0.4 release fo the new mathematical fonts:

> xma: The thorn in position 251 should be a lowercase thorn (having a 
> MC   descender)

Question: Could you perhaps provide some references showing examples
of the use of \eth and \thorn as math symbols?  Books, journals, papers?
Might be helpful, I suppose!

>      The incomplete differential (d with bar) is still missing. It should  
>      be included both as an upright and a math italic letter

Yes, I'd second that, in principle, remembering the use of \dbar and
also \deltabar in thermodynamics.  However, I'm afraid that the
tedious question of which upright/barred/slashed symbols to include 
or leave out in the MC encoding probably needs to readressed again
as a whole sooner or later.

Without wanting to reopen the discussion right now, here's a list 
of symbols that comes to mind to be decided upon eventually:

- upright letters, mathematical constants: 
  e (base of exponential function), i, j (imaginary unit)

- upright letters, differential operators: 
  d, \partial, \delta (variational), \Delta (increment),
  d with bar, \delta with bar, \partial with bar (same as \eth ?),
  D (differential along geodesics in general relativity)

- slashed or barred letters in physics:
  \hbar, \hslash (= h/2\pi in quantum mechanics), 
  \iotabar, \iotaslash (in fusion plasma physics),
  \lambdabar (do we need two shapes or not?), Vbar (?)

Possible criteria for inclusion/exclusion:
- availibility of ready-made symbols (e.g. barred letters)
- technical reasons, e.g. need for kerning with other letters
- inclusion in other encodings standards (ISO/SGML/Unicode)
- support or faked symbols in publisher's style files
- frequency of use and relative importance of fields

>      The upright lowercase greek letters need some adjustments

Yes, definitely.  For a start I'd suggest CTAN:fonts/greek/kelly,
which is very much like CM greek with minor adjustmenst needed
for use in upright shape.  On a longer term, however, I'd like to
remind you that it won't be enough to provide just one MC-encoded
font containing both upright/italic greek in medium/normal shape.  
We also need bold/bold italics, and bold sans/bold sans oblique.

> xmb: I am surprised by the two square root signs, which are _smaller_
> MX1  than the default one. Personally, I find them rather irritating.
>      How do the other members of this forum feel?

I think there is a technical reason for this.  The small version 
of the \sqrt radical (textstyle) is also used for \surd symbol 
(currently in cmsy10).  To be able to replace cmsy5/cmsy7/cmsy10 
by only one cmex10-replacement those three extra sizes are needed.
However, I'm afraid that they might be causing trouble about
the limit of 15 different depths/heights.

>      BTW, do the other mebers of this forum agree with the number and size 
>      steps of the growing delimeters? Are they really all needed (I am in 
>      doubt here)? An argument in favour of many sizes in one font could be, 
>      that one only needs one xmb font for all point sizes. On the other 
>      hand, TeX is not delimited by the absolute number of fonts loaded, but 
>      by the number of math font families. Thus, a package like xscale
>      is just fine. It also resolves the problems with the many different 
>      heights and depths.

I think I mentioned before that I'm somewhat skeptical about the
current proposal how to split MX1/MX2.  I'd rather go for a more
traditional MX font providing only the traditional range of sizes
and an optional MX1/MX2 combo providing extra sizes, but without
having to worry about compatiblitiy constraints concerning how
to distribute delimiters, bigops and wide accents.

BTW, I've got an unfinished implementation of such an MX encoding,
which includes the following:

0x00--0x27: miscellaneous: 2 reserved slots (0x00 and 0x20), 
            7 + 3 radicals, 7 or 8 horizontal brace pieces, 
            1 or 2 extension pieces for vertical braces,
            vertical arrows and bars, and a few empty slots

0x28--0x3F: 4x6 pieces for extensible delimiters 
            (parens, brackets, braces + semantic brackets (new))

0x40--0x5F: 4x8 pieces for big delimiters at four sizes each
            (parens, brackets, braces + semantic brackets (new))

0x60--0x7F: 4x8 pieces for big delimiters at four sizes each
            (angles, slash/backslash, floors, ceilings)

0x80--0xBF: 32x2 slots for big operators (including some new ones)

0xCF--0xFF: 8x8 slots for wide accents at eight widths each
  
> xmd: Johannes Kuester remarked recently, that the Vinogradov symbols
> MS1  should be differentiated from >> and <<. xmd seems the natural place 
>      where they could be added.

>      Another missing glyph: \VDash (\nVDash is in MS1)

>      Another missing glyph: \barvee

I recently came across some WWW pages of the W3C html-math containing
various references about proposed or already existing ISO/Unicode/SGML 
character/entitiy sets for encoding math symbols.  Does anyone have 
some spare time to compare those lists with our font tables to check 
whatever else we may have forgotten?  

http://www.w3c.org/pub/WWW/MarkUp/Math/
http://www.ams.org/html-math/

> xme: The scriptscriptstyle long arrows show visibile gaps in my printout and 
> MS2  on the screen. Probably the pieces should have better overlap.

I think I noticed another situation where some big delimiters didn't
overlap properly leaving a very tiny visible gap in xdvi, which wasn't
noticable on paper.  Perhaps some rounding problem?
 

Cheers, Ulrik.