Sat 17 Sep 2016 01:15:49 PM UTC, comment #10:
Ah. Now I remember better what that thing with the precomposed letters is about.
There are a couple of combinatinos, yod-patah and yod-hiruk, featuring high-placed vowel points, that are used in Yiddish. Only the yod-patah is marked as Yiddish in the Unicode standard, but I have been assured that in both cases, the Hebrew usage has the vowel placed low.
However, the font software plays a trick: it pre-composes many combinations into ligatures (my recollection is that it has not always been this way.) But it pre-composes these things regardless of the language of the text.
So Freefont had one table that de-composes these two ligatures, and re-composes them only if the text is marked as Yiddish. I'm putting that de-composing table back, but only for these two letters.
|
Wed 14 Sep 2016 03:37:57 PM UTC, comment #9:
Hi David,
see if current SVN isn't better.
I remember now. I de-compose pre-composed characters for Yiddish. Modern font software always pre-composes certain combinations, but it isn't always the right thing to do -- in this case, the Yiddish convention is to place the marks differently than appear in the pre-composed versions.
And I guess at the same time I decided to decompose all the Hebrew pre-composed letters -- which also wasn't the best choice.
Thanks!
|
Mon 12 Sep 2016 06:59:36 PM UTC, comment #8:
Hi David,
Generally, it is helpful to see examples. I make up examples of my own, but I'm just guessing.
Yes it could be that in the past, I thought it best to de-composed the pre-composed characters. I'll look at it tomorrow.
|
Mon 12 Sep 2016 06:52:39 PM UTC, comment #7:
Even with the new 'mset' table, the precomposed and decomposed vav-dagesh behave identically in FreeSans, i.e. the dagesh leaks into the next character’s space. For example, the word “וּפִי” looks wrong. The same goes for yod-dagesh. I notice that vavdageshhebrew has:
Could the MultipleSubs2 be interfering with the Ligature2?
|
Sun 11 Sep 2016 01:04:55 AM UTC, comment #6:
No, a dagesh would not actually appear on yod-yod or vav-vav.
|
Sat 10 Sep 2016 07:59:27 PM UTC, comment #5:
I added an 'mset' table meant to replace yod-dagesh and vav-dagesh with precomposed versons.
Tell me, does the dagesh really appear on yod-yod and vav-vav in text?
|
Wed 07 Sep 2016 10:07:29 PM UTC, comment #4:
Added mark positioning for the wide letters.
|
Tue 06 Sep 2016 08:12:57 PM UTC, comment #3:
It’s much better but not perfect.
In FreeSans, the dot on the thin letters (vav, zayin, yod, final nun, vav-vav, vav-yod, and yod-yod) should add some advance width: it is too close to the following letter.
In FreeSans and FreeMono, the dot on the wide presentation forms is still not placed correctly.
|
Mon 05 Sep 2016 05:21:37 PM UTC, comment #2:
Fix for all faces in SVN.
|
Sun 04 Sep 2016 06:56:46 PM UTC, comment #1:
See if FreeSerif isn't improved in the latest SVN.
|
Sun 04 Sep 2016 03:04:55 PM UTC, original submission:
All three fonts have problems with the Hebrew point dagesh.
In FreeSans, some glyphs are missing a vowel_left_heb anchor:
- yodhebrew
- finalnunhebrew
- tsadifinal
- vavvavhebrew
- vavyodhebrew
- yodyodhebrew
- yodyodpatahhebrew
- uniFB21..uniFB28
In FreeSerif:
- alefhebrew
- finalnunhebr
- finaltsadihebrew
- vavyodhebrew
- yodyodhebrew
- uniFB21
- uniFB22
In FreeMono, there is no vowel_left_heb anchor, so the dagesh looks wrong on nearly every letter, or at least different from the precomposed glyphs in Alphabetic Presentation Forms.
|