On 6/2/2019 8:36 AM, Mikael P. Sundqvist wrote:

On Mon, May 20, 2019 at 9:55 PM Wolfgang Schuster mailto:wolfgang.schuster.lists@gmail.com> wrote:

Mikael P. Sundqvist schrieb am 20.05.2019 um 21:41: > Hi, > > attached is the result of running latest lmtx on > > \definefontfeature[default][default][protrusion=quality,expansion=quality] \definefontfeature [default] [default] [threshold=20,expansion=quality] > \setupalign[hz] > \setupbodyfont[lucidaot] > > \starttext > Gauss öppnade med denna studie ett helt nytt fält inom matematiken, > det som > handlar om \emph{elliptiska funktioner} och \emph{modulära former}. > \stoptext > > What happens with the space between the characters in "studie", that > becomes "s tudie"? The spacing between some other characters are also > off (ma tematiken), but not as much as in studie. It's a problem with Lucida Opentype which lacks kerning (I reported the same problem a while ago).

Wolfgang

Hi,

thank you Wolfgang (I just realized I only sent my thank you to you privately)!

Your threshold=20 solved that particular case. But look at the following example:

\definefontfeature[default][default][threshold=20,protrusion=quality,expansion=quality,lnum=yes,itlc=yes]

\setupalign[hz,hanging,line,height]

\definebodyfontenvironment[9bp][interlinespace=12bp] \setupbodyfont[lucidaot,9bp]

\starttext Gränsvärde föreligger i en inre punkt i definitionsmängden om och endast om höger- och vänstergränsvärde existerar där och är lika. \stoptext

I attach the output (latest lmtx). The word "endast" shows up as "endas t".

So, what does this threshold do? I can play with the 20. If I change it to 5 it looks better here (not perfect), but my document is approx. 250 pages, so, I fear it will break somewhere else...

I hope that this does not mean that the Lucida font turns out to be useless :( Lucida does without kerning so where in other fonts we have more of less tight boundingboxes + kerns, in lucida we have large side bearings and in that case widening works out a bit different visually (thicker vertical strokes as usual with hz + expanded built in side bearings), definitely on screen.

Now, one thing that happens in the backend is that glyphs are output as sequences and scaled as whole and occasionally the starting points get synchronized .. this is what threshold deals with. Anyway, I might improve thinsg over time as in context lmtx we have a bit more control over matters. One can argue if intercharacter kerns have to be stretched in other cases but in context that is configureable. Hans ----------------------------------------------------------------- Hans Hagen | PRAGMA ADE Ridderstraat 27 | 8061 GH Hasselt | The Netherlands tel: 038 477 53 69 | www.pragma-ade.nl | www.pragma-pod.nl -----------------------------------------------------------------

On 6/2/19 12:47 PM, Hans Hagen wrote:

On 6/2/2019 8:36 AM, Mikael P. Sundqvist wrote:

...

I attach the output (latest lmtx). The word "endast" shows up as "endas t". that one is too weird indeed, might be a bug

I get weird results using TeX Gyre Pagella with expansion. I’m attaching a single page, typeset with LMTX (http://pdf.ousia.tk/lmtx.pdf) and MkIV (http://pdf.ousia.tk/mkiv.pdf) with a comparing file: http://pdf.ousia.tk/diff.pdf [I’m afraid the total size is too much for attachments to the list]). On the first line from the first paragraph, "pr" (from "proyecto") has broken kerning, "al" has also broken kerning. At least "el" has wrong kerning in the second line. On the third line, "de", "manera" (and I would say "están" have kerning issues. I think it is worth to print both pages and to check this on paper. I don’t know what changed, but probably a default that outputs similar text in LMTX to the one generated by MkIV would be great. I hope it helps, Pablo -- http://www.ousia.tk