Errors or ambiguities found in Specimen.tex

[Tchiller]

I am just checking out that fantastic font and would like to thank you in the first place.

Today I had a look at the Specimen document and would like to post some things, I stumbled onto. Sorry, it will be quite an unsorted list, but maybe some stuff is very easy to fix:

• you are loading +tlig. What is that. Not mentioned in the document. Maybe put a comment in the source
• 2.1 Locals: for english and german you could maybe link to the package selnolig. Your feature works but does seperate ligatures in too many cases
• 2.1 when separating the ligature fl (e.g. auflachen), the two letters touch each other
• 2.2 Character Composition and Decomposition: Some examples would be nice, especially as you load ccmp by default.
• 2.8 Contextual Alternates: does "should be on by default" mean, that you recommend it? I didn't get it when reading it first time. Sounds a bit like: "is switched on, if feature already works in this version". As you notice, I'm no native, so maybe just a "problem" for me
• 2.8 IRAQ - the R and A are touching each other
• 2.10.2 General Variants:
  • cv05: I can't see any differences. Please describe further!
  • cv06: Can't see any difference either. Maybe because you set cv06 in your preamble

OK, that's all I have noticed. Sorry for that messy list. But maybe it is some help. In my opinion, the Specimen.pdf is a great show case, so everything should be clear.

[Tchiller]

and one more small thing.

In your fourth last example paragraph, there are three chars not working in the last line.

{\scriptsize Чөлөөт нэвтэрхий толь, Википедиагаас}

[georgd]

Am 2014-02-08 10:48, schrieb Tchiller:

I am just checking out that fantastic font and would like to thank you in the first place.

Thank you!

Today I had a look at the Specimen document and would like to post some things, I stumbled onto. Sorry, it will be quite an unsorted list, but maybe some stuff is very easy to fix:

Thanks a lot for pointing all this out!

• you are loading +tlig. What is that. Not mentioned in the document. Maybe put a comment in the source

“tlig” or “Ligatures=TeX” is a pseudo feature which replicates some of tex’s behaviour in Xe/LuaTeX: ``'' -- --- ! ? << >> → “ ” – — ¡ ¿ » « I don’t need it any more (I input these characters directly) so I commented it out. In general I’ve added some comments in the source.

• 2.1 Locals: for english and german you could maybe link to the package selnolig. Your feature works but does seperate ligatures in too many cases

selnolig is a great package but only working together with LuaTeX. I’ll think about it. In which cases do you think the feature does too much?

• 2.1 when separating the ligature fl (e.g. auflachen), the two letters touch each other

I can’t reproduce this. Could you show me an example and the corresponding source code?

• 2.2 Character Composition and Decomposition: Some examples would be nice, especially as you load ccmp by default.

Ok, I’ll think about some.

• 2.6 Small Caps: Is there any possibility do give an option that suppresses accents on capitalized characters? I will investigate that problem but I remember, that it should not be done e.g. in spanish. For my taste, I would like to get rid of the accent automatically.

This is a font feature request. Iʹve thought about that already. It would mean either to decompose them and add alternate blank accent glyphs or adding alternate sc-glyphs. But I’d rather go with the first option so I can use the blank accent glyphs in UC too.

• 2.8 Contextual Alternates: does "should be on by default" mean, that you recommend it? I didn't get it when reading it first time. Sounds a bit like: "is switched on, if feature already works in this version". As you notice, I'm no native, so maybe just a "problem" for me

Ok, I’ll have to reword that. It means that the engine should enable the feature by default — which is a recommendation from the OT-spec but not mandatory. XeTeX (HarfBuzz) follows such recommendations but LuaTeX doesn’t in many cases (following a philosophy of only enabling whatʹs really necessary). “calt” is one of LuaTeX’s “victims”.

• 2.8 IRAQ - the R and A are touching each other

I’m fixing this in the font.

• 2.10.2 General Variants: - cv05: I can't see any differences. Please describe further!

That’s a bug in fontspec (https://github.com/wspr/fontspec/issues/115) which has been open for quite some time now. There’s a workaround which I applied for now, still hoping that the bug will be fixed some day.

• cv06: Can't see any difference either. Maybe because you set cv06 in your preamble

Well spottet. In fact, the difference is minute but

OK, that's all I have noticed. Sorry for that messy list. But maybe it is some help. In my opinion, the Specimen.pdf is a great show case, so everything should be clear.

Thank you!

and one more small thing.

In your fourth last example paragraph, there are three chars not working in the last line.

{\scriptsize Чөлөөт нэвтэрхий толь, Википедиагаас}

I forgot to add the Mongolian characters to the 8pt font.

[Tchiller]

Thanks for your quick answer and the already introduced fixes.

Everything works fine now and the documentation is much better. Only the Mongolian chars are still missing.

I found a new issue with kerning of the old s (shown in my MWE below).

Forget about the accents on capitalized chars. I already removed it from my post, when I found out that this was a wrong information on my side. Seems that in French and Spanish they should stay like they are. Maybe it was some vertical space issue like in Álvaro, Ángel, but for the smaller caps, this argument would not work. I decided to always write accents now (apart from that I never use caps for my own works...)

The ligatures and kerning things are shown below. Hope, you can reproduce it.

Thank you!

% arara: lualatex: { shell: yes, options: -synctex=-1 }

\documentclass{article} \usepackage{luatextra} \usepackage{fontspec} \defaultfontfeatures{% ,RawFeature={+dlig, +calt, +cv11, +ss05} ,Ligatures={TeX} } \setmainfont[% ,Path = C:/WINDOWS/Fonts/ ,Extension = .otf ,UprightFont = *12-Regular ,ItalicFont = *12-Italic ,BoldFont = *08-Regular ]{EB Garamond} \usepackage[final]{microtype} \usepackage{polyglossia} \usepackage[autostyle]{csquotes} \setdefaultlanguage[babelshorthands=true]{german} \setotherlanguages{english} \usepackage{showhyphens}

\begin{document} \enquote{f} und \enquote{ſ}

\textenglish{\enquote{f} and \enquote{ſ}

Examples from selnolig, where breaking the ligature would be wrong:}

{\addfontfeature{Language=German} Kafka Sognefjord Dovrefjell}

\textenglish{broken ligatures with touching heads:} auf"|lachen, auf/lachen \end{document}

[georgd]

Am 2014-02-11 13:23, schrieb Tchiller:

Everything works fine now and the documentation is much better. Only the Mongolian chars are still missing.

I know, I still have to fix that.

I found a new issue with kerning of the old s (shown in my MWE below).

Thanks, I’ll add these pairs

Forget about the accents on capitalized chars. I already removed it from my post, when I found out that this was a wrong information on my side. Seems that in French and Spanish they should stay like they are. Maybe it was some vertical space issue like in Álvaro, Ángel, but for the smaller caps, this argument would not work. I decided to always write accents now (apart from that I never use caps for my own works...)

The ligatures and kerning things are shown below. Hope, you can reproduce it. ‌‌… Examples from selnolig, where breaking the ligature would be wrong

There’s no case in German where breaking a ligature is wrong, there are no required ligatures in non-blackletter German typesetting. The ones you cite are exceptional cases where a ligature could be allowed between f and k or f and j (thereʹs no such tradition). However, I’ve provided such exceptions in the liga feature in https://github.com/georgd/EB-Garamond/commit/832f59c8391f557cd31921f5cb9d10acd69eb8a6#diff-8a354974f071f31b9b0517fc4b43df32. You can test with this font: http://georgduffner.at/ebgaramond/fonts/EBGaramond12-Regular.otf (the website usually features the current state of the fonts in the git repo)

\textenglish{broken ligatures with touching heads:} auf"|lachen, auf/lachen

That’s a TeX-specific problem. "| and / break the stream so Opentype features won’t be applied across the break point, neither kerning nor any contextual lookup, you’d have to apply kerning manually. In Xe- and LuaLaTeX I recommend to use the proper unicode symbol for breaking ligatures, the zero-width-non-joiner (ZWNJ), U+200C. Thus you’ll also get the alternate short-bowed form of f before l. Either input it directly (the new German standard keyboard layouts T2 and T3 have it on AltGr+.) or via \char"200C. I usually redefine "|: \defineshorthand{"|}{\char"200C} but I’m not so much a TeX-expert to say if that’s the correct way to do so.

[Tchiller]

You are incredibly fast. Thanks.

I was a bit quick with arguing on the German ligatures. When I did research on the "wrong" cases, I could hardly find some... So you are right here. No problem to use your local feature then (and the mentioned cases work with your new font)

Great Tip with U+200C. I had no Idea, that does exist. Works really great. I just have to find out, how to allow hyphenation at this very place.

Now everything looks great here. Thanks for your help and effort!

[georgd]

You’re welcome.

The hyphenation problem is something that’s probably to be solved by the babel team?

[Tchiller]

Another super small thing in the specimen. In the Serbian locale section, the second last letter to be replaced looks the same in both versions. Best regards.

[georgd]

This was originally a sha with a line below (parallel to t with line above) but at some point it was brought to my attention that this isn’t generally accepted in Serbian so I removed that substitution from the lookup but forgot to remove it here. I’ll move it to a separate feature.

[Tchiller]

No hurry, I don't need this char. Just stumbled on it on my search for Latvian symbols like ={i} in EB Garamond and noticed, that two symbols appear to be identical.

[skalyan91]

By the way, does the font have Bulgarian alternates? This would be easy to do, by repurposing Latin and IPA glyphs.

Sent with Mail Pilot

[georgd]

@Tchiller the Latvian ={i} (ī), does it have the i-dot too? I’ve read somewhere that in one of the baltic languages ī is constructed from the complete i, but I don’t remember which it was and haven’t found any further sources on that.

@skalyan91 I’ve thought about them. But they’d not come as a Bulgarian locale but as a styleset.

[Tchiller]

@georgd Sorry, I am no expert at all. Don't speak a word in Latvian. But I have never seen the i-dot here. Wikipedia (English, Latvian and German) does not use it. Google maps (look for Riga) does not use it. And typographically: I would leave the dot away.

[PanderMusubi]

In Specimen.tex it says that ligd is not switched on by default. However, it looks that it is and therefore needs an asterisk in the caption of \paragraph*{Discretionary Ligatures} Correct?

And the same goes for ligc.

[georgd]

@PanderMusubi, Iʹm not sure if I understand correctly. liga and clig should be enabled by default, but dlig and hlig shouldn’t, according to the specs. XeTeX does so, therefore if you look at the source, you’ll see that I had to use \addfontfeature in both cases.