Fulfilled or rejected wishes for Latin Modern

< Latin Modern | Wishes for Latin Modern >

This page contains the wishes from the page Wishes for Latin Modern that have already been fulfilled or rejected.

The content was moved from http://www.tug.org/tex-archive/info/lmodern/lmodern-done.html to this page with agreement with Harald Harders (h.harders@tu-bs.de) who kindly maintained that page during the most active years of developent of Latin Modern and was now looking for a new maintainer. Hopefully this wiki can offer a good-enough shelter for comments and proposals how to make the Latin Modern font family even better than it already is.

Fulfilled

Kerning

• Especially around quotation marks, see lm-kerning-086.pdf and lm-kerning.tex .

• Walter Schmidt : The font cork-lmri10 is lacking negative kerning between W-a. Please, compare with ecti1000 ! Note that I did not search systematically for further deficiencies of this kind.

• Walter Schmidt : Upright fonts should exhibit some negative kerning between e-V, with respect to the physical unit ‘eV’.

• Walter Schmidt : Jörg Knappen's EC fonts include the e-V kerning and other additional kerning data, which were not present in the old CM fonts. See the file exrligtb.mf. IMO, these additional data should be adopted for Latin Modern, too.

• Morten Høgholm : Kernings with ‘æ’ and ‘å’ should be improved (e.g., ‘Tæ’, ‘Tå’, ‘Væ’, ‘Vå’).

• Harald Harders : Spacing in ‘[…]’, ‘(…)’, and ‘{…}’ should be symmetric.

• Harry Schmidt (2003-11-26): The distance between the letters A and u in the roman, roman/bold, italic and italic/bold font shapes is too large; there should be some negative kerning. The same holds for the letters A and v. Example: ‘Aufgabe’

• Reinhardt Kotucha (2003-12-23): I hope that it isn't too late, but I'm not very happy with Knuth's ‘Au’ kerning in \rm . I just saw that Latin Modern has the same problem. Example: ‘ Auto ’. I didn't test \sf and \it .

• No kerning between the letters k and a in the roman and roman/bold font shapes.

• Jacko/Janusz (2004-04-16) : All kerning has been prepared from scratch. Previously, the starting point was AE family of fonts; the starting point for the current version of LM fonts were kerns of EC fonts. Extending them properly so that all necessary diacritical characters are kerned was the most tedious part of this stage of works on LMs. Of course, we exploited several programistic tools prepared especially for this occasion (AWK scripts), as altogether there are nearly 300000 kern pairs, circa 6000 per font ;-)

Single glyphs/ligatures

• Harald Harders : The slash (‘/’), backslash (‘\’), and plus (‘+’) are much nicer than in European Computer Modern.

• Harald Harders : The guillemets are nicer than in European Computer Modern.

• Gerrit Kirpal : The text fractions are better because the fraction line is a solidus instead of a horizontal line.

• Haruhiko Okumura : \textyen (Japanese currency symbol) has a too small ‘equals’ sign. It would be much better to just say \def\textyen{Y\llap=}, or somewhere in between.
  • Jacko/Janusz (2004-04-16) : We attempted to improve it.

• Uwe Stoehr (2003-12-10): The \textcircled is small with Latin Modern : textcircled.tex , textcircled.pdf
  • Jacko/Janusz (2004-04-16) : Not touched yet.
  • Harald Harders : Recent versions of textcomp do not have this problem.

Rejected

Kerning

• Thilo Barth : Kerning between d and a should be closer ( his German news posting ).
  • Harald Harders : I don't agree.
  • Gerrit Kirpal does not agree either.

Single glyphs/ligatures

• Stefan Nobis : The horizontal strokes of the ‘[’ and ‘]’ should be a little bit longer.
  • Jacko/Janusz (2004-04-16) : Not sure... This is the feature inherited from CM fonts. We are rather reluctant with respect to the changing of design.

• Hans Hagen (2004-01-25): Example in Latin Modern Mono:

‘cigarettes per day --- and we humans’.

Looks like cork-cmtt10 has a strange ligature since texnansi-cmtt10 does ist right. In cork-lmtt10 I find the following lig:

(LABEL 0 55)
(LIG O 55 O 25)
(LIG O 177 O 177)
(STOP)

This gives a funny character an ddefinitely no ‘ --- ’. I wonder, shouldn't the tt tfm's have no ligatures at all? (vf's may have them).

• Jacko/Janusz (2004-04-16) : This _is_ the proper behaviour. If you disassemble ectt1000 you will see the same figures. The complete set of ectt1000 ligatures is as follows:

quoteleft + quoteleft --> quotedblleft
quoteright + quoteright --> quotedblright
hyphen + hyphen --> endash
hyphen + hyphenchar --> hyphenchar
less + less --> guillemotleft
greater + greater --> guillemotright
comma + comma --> quotedblbase
exclam + quoteleft --> exclamdown
question + quoteleft --> questiondown

Ask Jörg Knappen about the idea behind this ;-)

• Gerrit Kirpal : Latin Modern should be darker.
  • Harald Harders : I am not sure. The fonts could be slightly darker, but not much. They still have to fit to the mathematical type 1 fonts which makes a darker lm font problematic.
  • Boguslaw Jackowski: Subtle darkening is feasible for bitmap fonts; with outline ones it is nearly impossible. Moreover, I agree with you that they must fit with math.

General things

• Sometimes, I am missing the bold and the bold italic typewriter font shape.
  • Jacko/Janusz (2004-04-16) : Once upon time Jacko tried to prepare a bold version of cmtt - and failed. It seems that Knuth's design precludes (acceptably neat) bold typewriter. The resort would be using courier-like fonts.
    Moreover, even if we succeeded in setting parameters properly, the whole font should be prepared from scratch (remember that the groundwork LM fonts were AMS/BlueSky Type 1 fonts for CM family). It would require much work, actually, too much (for us, at present).

• There should be an italic typewriter font with the same character width as the upright shape.
  • Jacko/Janusz (2004-04-16) : There is an italic typewriter font: lmtti10 (a counterpart to lmtt10). Adding lmtti8 and lmtti9 with the metric corresponding to lmtt8 and lmtt9, respectively, would mean -- in our opinion -- an effort not worthy of the result. Using lmtti10 at 8pt and lmtt10 at 8pt, if needed, yields -- again in our opinion -- satisfactory results.