notofonts / syriac

Noto Syriac
SIL Open Font License 1.1
4 stars 1 forks source link

Assyrian Syriac fonts not found #10

Open r12a opened 4 years ago

r12a commented 4 years ago

I'm unable to find the following Syriac fonts in this repo:

NotoSansSyriacEastern NotoSansSyriacEstrangela NotoSansSyriacWestern

There's only a Noto Sans Syriac font. The differences between these various fonts are significant and essential for writing modern Assyrian text.

Looks like they are available still from https://www.google.com/get/noto/. Should they not also be here ?

dscorbett commented 4 years ago

Noto Sans Syriac combines all three styles in one font. The language system tags SYRJ and SYRN enable the non-Estrangela styles. This was clearly a deliberate design decision, but since the glyph sets are so distinct it would be better to split them into three fonts again. The current situation is like merging Noto Nastaliq Urdu into Noto Sans Arabic as a stylistic set.

r12a commented 4 years ago

It's also not much help if the language tags you're using are aii and tru, which is what i used when i tested for that possibility earlier. :(

dscorbett commented 4 years ago

If certain Syriac languages are always written in certain script styles, could you submit feedback on the OpenType language system tag list explaining which BCP 47 tags should map to SYRE, SYRJ, and SYRN?

punchcutter commented 4 years ago

Merging these was one of the dumbest things I've done. I argued against it, but was told that Google wanted it this way so we had to do it. I said ok I'll do it even though it will make the fonts unusable in most situations. They were 3 separate fonts from the start and merging them was actually a huge pain for no real benefit.

marekjez86 commented 4 years ago

@punchcutter : Thank you. I also see no reason why we did this. This was unnecessary -- and I think Google wasted its money and your time. Did we make any changes in the fonts since the time we had three separate ones? I'd love to just go back to this situation.

punchcutter commented 4 years ago

@marekjez86 I believe we did it all in one chunk so there were major updates to the 3 separate files and before delivering them as separate updates we merged them into one. I'm pretty sure I built the 3 as separate projects first (anticipating this) and then did the merge, but all those pieces are at Monotype so I don't know anymore.

simoncozens commented 1 year ago

Currently we have the weights in separate sources (because of layout differences between them) and multiple different styles in each source. It's completely the wrong way around. I'm going to try unpicking these, but I can predict it's going to get messy. I'll try my best to keep things compatible, and use regression tests to help make sure most of the behaviour is the same, but still - apologies if things get broken along the way, but somehow we have to make progress...

simoncozens commented 1 year ago

OK, I managed to untangle the three variants, and as a bonus made them into variable fonts along the way. The draft new fonts are here (zip file download). Testers wanted!

simoncozens commented 1 year ago

I fixed a number of positioning and kerning issues since that draft, and have used Marc's wonderful diffenator2 to ensure that the new fonts behave identically to the old fonts, at least on this word list I generated from the Peshitta. I want to clean up some of the layout rules and then I'll do a new release.

r12a commented 1 year ago

I did some informal testing against https://r12a.github.io/scripts/syrc/aii.html and https://r12a.github.io/scripts/syrc/tru.html (and set them to use webfonts derived from the original Noto Sans Syriac Eastern font). Mostly, looks good, but i found the following. Images are screenshots of the page with the old font first, then the page with the new.

[1] the new version of the font doesn't handle tatweel properly. See https://r12a.github.io/scripts/syrc/aii.html#cursive_shaping and https://r12a.github.io/scripts/syrc/tru.html#cursive_shaping. The older webfont produced nice connections, but the new font appears not to connect.

Screenshot 2023-01-03 at 17 28 12 Screenshot 2023-01-03 at 17 28 26

[2] See the example in the text just above figure 22 of the Eastern page https://r12a.github.io/scripts/syrc/aii.html#fig_alaph_ligature_l. The ligature for aleph after lam isn't produced in the new font.

Screenshot 2023-01-03 at 17 16 34 Screenshot 2023-01-03 at 17 18 16

[3] In that same example, the high diacritic on the initial letter has fallen down a bit. Also, the diacritics have lost the old rounded shape and become just lines (which i don't think is correct - there are many diacritics in Syriac and dots are different to lines). These differences appear throughout the page.

[4] In the same paragraph, another ligature no longer forms (at the end of the word).

Screenshot 2023-01-03 at 17 25 32 Screenshot 2023-01-03 at 17 25 48

[5] In the Turoyo (western) page there's an example of SHIN in the first large orange word, which appears to be smudged in the new font.

Screenshot 2023-01-03 at 17 34 13 Screenshot 2023-01-03 at 17 34 56

[6] At figure 17 of the assyrian (eastern) page the 2 dots no longer appear side by side. (And the shape of the letter is more rounded than previously.)

Screenshot 2023-01-03 at 17 41 08 Screenshot 2023-01-03 at 17 41 01

[7] In the example just after figure 20, the positioning of the diacritics below the letters also seems wrong in the new font, and the spacing between the first and second letters has also changed.

Screenshot 2023-01-03 at 17 44 33 Screenshot 2023-01-03 at 17 44 41

[8] In the section on plurals https://r12a.github.io/scripts/syrc/aii.html#syame the positioning of the diacritic that places the dots above and below the letter is wrong.

Screenshot 2023-01-03 at 17 48 55 Screenshot 2023-01-03 at 17 49 05

[9] There are, of course, a bunch of things that the font could do better (such as support for ligatures) but i guess that's future work. Note, however, that it doesn't appear to have fixed https://github.com/notofonts/syriac/issues/8, even though you closed that issue. I'm still seeing 3 dots.

Screenshot 2023-01-03 at 16 38 25

I checked the Western font in the same way, and this time found that the tatweel was NOT supported (whereas with the webfont it is). See https://r12a.github.io/scripts/syriac/tru.html#cursive_shaping

[10] I'll stop there, in case it's just that i've done something wrong when installing the new fonts. I'm doing these check on a Mac using Firefox. I haven't checked the estrangelo version of the font. There used to be a font named Noto Sans Syriac Estrangelo, as i remember - is that now replaced by the Noto Sans Syriac font?

hope that helps

simoncozens commented 1 year ago

This is brilliant, thank you. I'll get this fixed...

There used to be a font named Noto Sans Syriac Estrangelo, as i remember - is that now replaced by the Noto Sans Syriac font?

Yes, the Estrangela style was the default for Noto Sans Syriac (with the other styles as language overrides), so I've kept that and split the other styles out as their own fonts.

simoncozens commented 1 year ago

I've tried to replicate your tests, and there are a variety of things going on here, some of which are related to my reworking and some of which are pre-existing issues...

1) I can't replicate the tatweel thing. I've checked in Harfbuzz and in the browser and it connects just fine. 2) Confirmed, ܠܲܝܠܹܐ is not forming a lamad/alaph ligature in Eastern. That seems to be a new problem with 3.0. 3) Confirmed, the 0732 glyph should be split up into an above/below pair, and it's not being. However, the lack of nice round dots is a pre-existing issue; dots were rounded in 1.04 but not in 2.001. 4) Confirmed, taw/alaph ligature not being formed in Eastern. This is also a new problem. 5) This is an old problem. That rounded form of initial shin (uni072B.Init) in Western doesn't seem to appear any more in the 2.001 sources, so we can't use it. 6) The shape change of dalath ( ܕ݂ ) in Eastern was made in 2.000, so not a new issue. But the position of the rukkakha is a new problem. 7) This is the same problem as (3). I've since fixed the kerning issue. 8) Same problem as (3). 9) This definitely works for me! I'm going to add shaping tests for it, too.

r12a commented 1 year ago
  1. I can't replicate the tatweel thing. I've checked in Harfbuzz and in the browser and it connects just fine.

Grr. Today it's also working fine for me. Sorry for causing you more work. Not sure what changed...

  1. The shape change of dalath ( ܕ݂ ) in Eastern was made in 2.000, so not a new issue. But the position of the rukkakha is a new problem.

Fwiw, the rukkakha seems to move as expected in the western font, but not the eastern one.

Screenshot 2023-01-06 at 12 50 39 Screenshot 2023-01-06 at 12 50 29
r12a commented 1 year ago

Btw, to avoid possible confusion i should mention that i edited the links in my earlier comment. I'm in the process of moving the orthography description pages to new locations (and a new template version). So the latest versions of the links above such as https://r12a.github.io/scripts/syriac/aii.html#syame are now https://r12a.github.io/scripts/syrc/aii.html#syame . (syriac -> syrc)

Over the coming days/weeks i'll be removing the files from the old directories and adding redirects.

ledzeppelin commented 3 months ago

@simoncozens i dropped a note here https://github.com/notofonts/syriac/issues/2#issuecomment-2297716762 about the increased use of Noto Sans Syriac Eastern

is there any way we can prioritize improving the placement of the rukkakha?

also im seeing the issue in examples like ܫܵܟ݂ܹܪ when using Noto Sans Syriac Eastern