Am Wed, 13 Aug 2008 23:07:43 +0200 schrieb Arthur Reutenauer:

Hello Ulrike,

...

1. In the following document there is no space between the a and the y:

\documentclass{article} \begin{document} \noindent a\\y

\bigskip

a\par y \end{document}

It looks as if the lines have only the depth from their content.

I don't seem to see that (see attachment). How did you generate the format? I used “luatex -ini pdflualatex.ini” where pdflualatex.ini contains

\pdfoutput1 \input latex.ltx

I tried it with a format generated by simply inputting latex.ltx (and then \pdfoutput=1 in the document), with one which loads also pdftexconfig.tex, and with a plain one: % Thomas Esser, 1998. public domain. % hyphenation tables are now set up in language.dat \input pdftexconfig.tex \input plain \dump \endinput In all cases (also if I try dvi-output) the lines glue together. The attachment text-lua is the pdf of the latex try.

...

2. I found here some code to use truetype and opentype fonts: http://luatex.bluwiki.com/go/Use_a_TrueType_font

Oh, that ;-) You really have to realize that these lines of code are nothing more than an experiment I wrote one and a half year ago, and you should be prepared to extend it if you want complete TrueType support (not to mention OpenType features, which are absent altogether).

Yes I realized that ;-). Currently I'm trying simple code to understand the principles so that I can later steal the right things from the context files.

That said, the problems should not be as bad as the ones you report below:

...

a) I tested it with a latin modern font and arial and with both fonts all lines of a text are printed one above the other. of the lines have a height and a depth.

That's not what happens to me, you should really send a precise example.

See test-lua2.tex + pdf

...

b) I also tried MnSymbol5.otf and there I get an error "invalid character". (I tried to get an "A"). With xetex the font works fine.

I can't confirm that. I get the A all right (see attachment). Again, please send the precise example that fails.

The MnSymbol-Bold5 you used works fine for me too. MnSymbol5 is the font doesn't work (the code is -- commented -- in test-lua2): Fatal error occurred, no output PDF file produced! -- Ulrike Fischer

Wolfgang Schuster wrote:

...

In all cases (also if I try dvi-output) the lines glue together. The attachment text-lua is the pdf of the latex try.

you can make a plain format operating in the same 'space' as context with luatools --make plain you then can process files with luatools --fmt=plain test eventually i will make a plain variant (probably a bit more more plain than plain) with basic open type support (for other formats you might need to add --compile to the make call) Hans ----------------------------------------------------------------- Hans Hagen | PRAGMA ADE Ridderstraat 27 | 8061 GH Hasselt | The Netherlands tel: 038 477 53 69 | fax: 038 477 53 74 | www.pragma-ade.com | www.pragma-pod.nl -----------------------------------------------------------------

Ulrike Fischer wrote:

But what would be really fine were extracts of the context code: E.g. a bundle of self contained lua-code and tex-code that leds to an extended \font command (or a \font command with extended syntax similar to xetex) with opentype/ttf-support. If such a primitive were present it would be probably quite easy to write the LaTeX support e.g. by adapting fontspec. Or a bundle which can be used to convert other input encodings to utf8. Or a bundle of the code which sets catcodes and lccode etc.

wel, if there is some kind of plain mode, then using that other macro packages is possible too, but it has no high priority; first most of the font stuff needs to be in place, then i start working on a more layered context setup (so that one can use components for other purposes) Hans ----------------------------------------------------------------- Hans Hagen | PRAGMA ADE Ridderstraat 27 | 8061 GH Hasselt | The Netherlands tel: 038 477 53 69 | fax: 038 477 53 74 | www.pragma-ade.com | www.pragma-pod.nl -----------------------------------------------------------------

On Thu, Aug 14, 2008 at 11:07 AM, Ulrike Fischer wrote:

Am Thu, 14 Aug 2008 09:58:14 +0200 schrieb Wolfgang Schuster:

...

...

...

...

a) I tested it with a latin modern font and arial and with both fonts all lines of a text are printed one above the other.

...

...

...

That's not what happens to me, you should really send a precise example.

See test-lua2.tex + pdf

This was a bug, you need a newer luaTeX.

http://www.mail-archive.com/dev-luatex@ntg.nl/msg00033.html

Ah. Fine. Then I will have to wait until a new binary for win is available.

You could use the binary from Akiras site. You need the dev-version, look at the bottom. http://www.fsci.fuk.kindai.ac.jp/kakuto/win32-ptex/web2c75-e.html Wolfgang

Perfect. With this version both spacing problems are solved.

Glad to hear that. But the problem with MnSymbol is still there, isn't it? I can reproduce it using MnSymbol5.otf with my code and -- what's more worrying -- with ConTeXt itself. I'm not sure what's at fault, LuaTeX or the font. Arthur

Ulrike Fischer wrote:

Am Thu, 14 Aug 2008 15:49:54 +0200 schrieb Arthur Reutenauer:

...

...

Perfect. With this version both spacing problems are solved. Glad to hear that. But the problem with MnSymbol is still there, isn't it?

Yes.

...

I can reproduce it using MnSymbol5.otf with my code and -- what's more worrying -- with ConTeXt itself.

I don't find this very worrying: Bugs which can be reproduced on common systems can be solved more easily. ;-)

taco and i spent quite some time today in testing lots of fonts, and we think that in the latest beta locating by name works ok; however, there are so many weird fonts out there that we probably need to spend a week figuring out what to do with them .. be tolerant or just reject them; currently our tests with many (also very huge) fonts run into problems due to memory usage when finalizing the pdf file (we're talking gig's of memory here) so that's another issue we will try to address (sometime after the context conference) in any case in mkiv we will have some mechanism where certain fonts are rejected from the name database (one can still load them by file: but at least they dont get reported then by --list); think of those cm super fonts Hans ----------------------------------------------------------------- Hans Hagen | PRAGMA ADE Ridderstraat 27 | 8061 GH Hasselt | The Netherlands tel: 038 477 53 69 | fax: 038 477 53 74 | www.pragma-ade.com | www.pragma-pod.nl -----------------------------------------------------------------

Charles P. Schaum wrote:

Hi all,

Quite a while ago I mentioned letter spacing, or the normal absence thereof, in TeX and friends. True, you could skin this cat by mathematically scaling the text (font size) by steps to fit the box --- the code is out there. But that was not the only way, thought I. Some programs also call this general idea tracking, that is, inserting positive or negative kerns independent of the \spacingfactor.

Just for grins I nabbed two snippets from, respectively, manmac.tex and texbook.tex. I then added a bit of my own hackery that was inspired by DEK's length macro. I realized that a non-counting version of the length macro was a memory-friendly way to insert a symbol between any two others in a given sequence. What I wanted to insert was a negative kern.

The negative kern must be small, less than \negthinspace, or else you start getting weird effects like a practically chiastic ordering of letters, and certainly a chaotic one. The smallest that one dare go seems to be -.066667em, the point where the serifs can start running together. Using -.033333em really does not show any letters running together, yet it still saves space. I got those figures as suggested by Knuth's own selections for tightness in the interword spacing.

And yes, you could theoretically space out the letters, like the soul package that does proper Sperrdruck. One major drawback with my kludge is the need to use control spaces, else all spaces evaporate. If soul were ported to ConTeXt, and it seems like the code might allow that, then a version that "spaces in" and "spaces out" could well be possible.

in my opinion letterspacing only makes sense in titles, not in the main body of text (i consider kerning to be part of the font design) anyhow, since we have to letterspace titles and such, there is provision for it in the core (non advertised low level stuff) \stretched {The badness of this line is very bad.} \hbox spread 2cm{\stretched {The badness of this line is very bad.}} \def\stretchedspaceamount{.25em} \stretchednormalcase {The badness of this line is very bad.} \def\stretchedspaceamount{.5em} \stretchednormalcase {The badness of this line is very bad.} \def\stretchedspaceamount{-.125em} \def\stretchedspacefactor{-4} \stretchednormalcase {The badness of this line is very bad.} \hbox{The badness of this line is very bad.} ps. the negative kerning was not supported, but i enabled it in the beta ps. in mkiv there is another model, one that permits kerning of arbitrary text Hans ----------------------------------------------------------------- Hans Hagen | PRAGMA ADE Ridderstraat 27 | 8061 GH Hasselt | The Netherlands tel: 038 477 53 69 | fax: 038 477 53 74 | www.pragma-ade.com | www.pragma-pod.nl -----------------------------------------------------------------

A while ago I mentioned a ``Problems after [n] pages'' issue in AUCTeX when using ConTeXt with it. Recently I used LaTeX and, hey presto, I was able to reproduce the same message. I ran LaTeX, then re-ran it to get the references right. I then ran it again and got a ``Problems after [0] pages'' message. But there were no problems. I ran LaTeX again, and this time it formatted 220 pages successfully. I really don't know the cause of the message, but there may be some kind of corner case. At least you know that it probably is not ConTeXt that is the source. Either way, I can live with the situation as long as my output is correct. Charles