Fri 28 Oct 2016 08:12:04 AM UTC, comment #12:
Proposed policy for ij ligature in Dutch:
- The ligature is formed by default.
- When displayed with extra space between letters, the ij is not expanded.
- When accents are put over the i and j together, they still behave as single gWlyph.
- fi does not ligate if followed by j
To make the ij not form (a in e.g. "Fiji"), there are two options:
1) mark the text language as not being Dutch
2) insert a ZWNJ character between the i and j
Note: the ij is really ligated in serif italic, and in the sans bold faces, there is a special glyph for IJ.
(file #38824, file #38825)
|
Thu 13 Oct 2016 12:23:25 PM UTC, comment #11:
That blog post only says that Dutch keeps the dot of the ⟨i⟩ in the sequence ⟨fij⟩. It doesn’t say that Dutch avoids all ⟨fi⟩ ligatures.
|
Thu 13 Oct 2016 07:38:56 AM UTC, comment #10:
Well, for German and Turkish already the ffi isn't default. Maybe I'll add Dutch to that list. It's messy business. (Compare with the "long s", which can also consume a lot of time.)
|
Wed 12 Oct 2016 10:48:08 PM UTC, comment #9:
I just came across this blog entry from a Dutch foundry
http://www.underware.nl/blog/2014/10/typesetting-the-dutch-ij/
There's not much extra information about the ij + acute issue, but there is a new (to me) issue about how the fi and ffi ligatures should combine with ij.
|
Tue 11 Oct 2016 02:41:44 PM UTC, comment #8:
A further complication is that, a few days ago on the Unicode mail list, it was pointed out that the lowercase ij ligature U+0133 officially does not have the "soft dotted" property.
It's all quite a muddle right now. With Michael Everson involved in the discussion, I wouldn't be surprised if next year's update to Unicode will clear this up.
(I was typing this as your message came through)
|
Tue 11 Oct 2016 02:32:25 PM UTC, comment #7:
OK, for text marked as Dutch language, the sequence
i - acute.cmb - j - acute.cmb
now produces a ligature with two acutes above the ij.
|
Mon 10 Oct 2016 06:22:03 PM UTC, comment #6:
See attachment ij-italic.png
(file #38696)
|
Mon 10 Oct 2016 12:33:29 PM UTC, comment #5:
A further thought: I made the ij ligature standard only for text marked Dutch. So far as I know, in Dutch, this is the only application of that combination.
(Am I wrong about that? I seem to remember somebody asserting that it was the case, and doing some research...)
As it stands, in italic anyway, there is no way to put the two accents on the ligature.
But I could make a further replacement of the sequence i acute j acute to a special ligature¸ again only for Dutch. This doesn't fix the issue of putting an accent directly on the precomposed ligature, but it provides a way of producing the accented ligature, that is at least not worse than automatically making the ligature itself.
|
Mon 10 Oct 2016 08:18:27 AM UTC, comment #4:
The case for a special measure is stronger for italic, where the letters are actuall ligated. There, using an accent on each letter breaks the ligature.
I'm still unwilling to implement encodings that remain controversial.
|
Thu 06 Oct 2016 02:15:31 AM UTC, comment #3:
Auto-doubling the acute accent on these digraphs is so unlike the behavior of any other letter–diacritic combination in the Latin script that I would be wary of implementing it until Unicode explicitly condones it.
|
Tue 04 Oct 2016 06:43:07 AM UTC, comment #2:
As for whether the IJ character "shouldn't be used", it certainly isn't deprecated the way that, for example, U+0149 ʼn is. The Unicode Standard text says "a vast amount of Dutch data is encoded without this digraph character" while pointing out advantages to using the ligature. I'd say it is uncommon but perfectly legitimate for people to use.
It seems within reason that someone would want to use 0132,0301 or 0133,0301, so it would be a good idea for a font to support them.
|
Sat 01 Oct 2016 05:43:24 PM UTC, comment #1:
Hi David,
I would be happy to implement soemthing like this if that is standard behavior. But in the discussion near as I can tell, there are a couple of opinions how to implement this, as well as several expressions of the opinion that the precomposed form was mistakenly encoded and shouldn't be used at all.
If the latter is the case, the solution is already there: put an accent over each the i and j.
Otherwise, I am indifferent to which of the proposal with accented letters joined with a ZWNJ or the one with precomposed letter followed by an accent. I'm not keen on implementing both, for something that might not be a good idea anyway.
Please let me know if I've misunderstood anything here.
BY THE WAY this reminded me of a longstanding issue with replacing the dot over i and j with an accent. There was an implementation years ago, but it's been broken...
Well now it's fixed in SVN.
Thanks for pointing this out!
|
Fri 30 Sep 2016 11:40:40 AM UTC, original submission:
There's a current thread on the Unicode mailing list that's relevant to font design
http://www.unicode.org/mail-arch/unicode-ml/y2016-m09/0086.html
It involves using the combining acute accent U+0301 with the IJ ligatures U+0132, U+0133. They say the correct solution is to add acutes to both the I and J using a single 0301.
A free typeface called Brill (http://www.brill.com/about/brill-fonts) does this; I have attached a clip from its user guide. This special rule is also used by Adobe's Source Code Pro font (but not Source Sans Pro).
|