[NTG-context] Fallback fails for Linux Libertine O to Junicode over private area, debugging?

Hans Hagen j.hagen at xs4all.nl
Sat Jan 8 11:42:20 CET 2022

On 1/7/2022 10:39 PM, J. P. Ascher wrote:
>> in that case you can best use a rather bare font i.e. no features at
>> all apart from kerning and use dynamic features grouped
>> in lmtx there are other tricks too
>> \noleftligaturing
>> \norightligaturing
>> \noligaturing
>> and other conbtrols like
>> \noleftkerning
>> etc
> This is very useful.  I'm thinking of writing an environment that
> switches: I like keeping the ease of typing normal characters, rather
> than getting glyphs directly, because it's easier to proof on screen in
> the editor.
>>> writing about reproduced in my work and thus now can in the historical
>>> item itself.
>> one aspect is how ligatures are made: some fonts have single shapes,
>> others use replacements and kerning of (then) multiple shapes
> I think you wrote about this somewhere else and it got me interested in
> physical type from my period: it looks like some printers did some of
> the same things, sometimes!  You might have a box of pre-composed
> ligatures, or not; you might take a knife to some letters to tweak them
> for specific uses, or not; you might even shave wrong-sized letters to
> fit them in where they don't belong.  Those shaved down letters could
> end up back with their siblings, now shifting around during printing, or
> they could be tossed.
> The pre-composed ligatures might most often be made on matrices struck
> with the same punches as for the non-ligatured glyphs, but typefounders
> might re-strike matries now and then, so you might end up with slightly
> different pre-composed ligatures in different, or even the same, fount
> of type.  It's nuts!
> Fred Smeijers writes about this a little bit in *Counterpunch*, but
> I had already started thinking about it because of ConTeXt.
> I wouldn't have known to look if it weren't for the typographical
> education you and your colleagues pass around; so, thank you again!
One problem with writing about things related to fonts is that when a 
font changes rendering changes 9this is a pain for manuals or articles 
that then no longer show examplex right)

for that reason one needs to

- save a font, best rename it
- define fonts by filename
- make sure that font specific manipulations are also saved

Fonts more and more have become like programs. In the past a font never 
changed (even expensive fonts had no update policy like programs do) but 
with open type fonts became way less stable especially free ones 
(vendors still freeze and then just sell them). Add to that forking, 
abandoning etc. With complex features being used, and features not 
always being applied consistently (bascially we have 1-1, 1-many, 
many-1, many-many mapping plus contextual lookups) we can expect bugs 
and they get fixed or not. It also depends on views of designer, 
programs being used to make them, programs used to test them, all of 
which can change over time, so in a way fonts have become less stable (a 
bit like hyphenation patterns as more fluid resource). At some point we 
can expect bug(let)s or imperfections to stay and adapt to that (as we 
do with math now). (This is not new: for a long time tex users had to 
deal with computer modern fonts with basically only proper kerning for 

Add to that evolving technologies (color fonts, image based ones e.g. 
emoji, variable fonts) that show up, then need fonts, where often the 
first ones are not okay, so bug show up, features changes, specs become 
better (or worse), bugs become features when programs compensate for 
issues, often a side effect of release before maturing (the haste of/on 
the web) but all that can settle after a decade (because in the tex 
world we talk decades anyway). In context so far we managed to keep up 
with e.g. color fonts, variables fonts pretty early (we were one of the 
first to actually typeset with variable fonts) but that of course also 
comes a price of later adaptation (when fonts show up, specs become 
clear). Here definitely a longer time scale matters so using such fonts 
in a project is more tricky for various reasons.

The best would be if we had a repository for context where we put fonts 
(clean file names, maybe with normalized version/date in file name) that 
we can then trust to stay the same. In that case a typescript and lua 
extensions can check for versions reliable.

(We might do that for math to start with.)


                                           Hans Hagen | PRAGMA ADE
               Ridderstraat 27 | 8061 GH Hasselt | The Netherlands
        tel: 038 477 53 69 | www.pragma-ade.nl | www.pragma-pod.nl

More information about the ntg-context mailing list