Abhaya Libre Font (Sinhala, Tamil, Latin)

[Pathum Egodawatta]

Hi All,

We are working on refactoring 'Abhaya' a classic Sinhala font from designer Pushpanada Ekanayake(Sri Lanka) Abhaya is the most printed Sinhala font at the moment in Sri Lanka. A high contrast modern font that originally had two weights (Regular, Bold). Designed in 1997 and then pirated all over it is the de-facto font for novels, most of the newspapers, documents. Most universities require their students to submit written essays in 12pt Abhaya. See the paring-feedback.pdf  for a sample and and this folder for some sample scans.

Sol Matas is working on a new Latin set and we are using two Shree Lipi Tamil fonts to match with Abhaya. The most of the work is done by Pushpananda and I'm helping him to coordinate and work with all of us contributing.

Currently we are three weeks into the project, If you would like to see the current state, just jump to my next post. Or checkout the GitHub

Im going to quickly sum up our development so far.

1. We started off looking at obvious things to fix on old Abhaya, as many old ASCII era dtp fonts,

1. It was in its own proprietary encoding.

2. Drawings  from 1998, with crooked outlines here and there (the CRT effect, as Pushpa put it)

3. So many points

2. After initial discussion, Sol, sent us some samples with different fonts that she thought that could match Abhaya and Tamil we had.We sent her feedback, What we thought could work. Was a great exercise to understand the characteristics that made up Abhaya.( paring-feedback.pdf )

1. Abhaya Latin needs a high x height.

2. With tamil being more Bodoni like and Abhaya being  more classic, Sol suggested to tweak the Tamil a bit to match the feeling of Abhaya. As trying to make a latin that is in between would be useless.

3. And we sent Sol some documents and newspaper scans so she could get a better feel of Abhaya in context, and print. Also so blogs that she can see Sinhala and Tamil in use.

4. And we thought we could follow some of the height matrix made by Rafael Sarvia when designing his ‘Serendip Sinhala’ font.

3. Then based on our feedback she did the first draft of Latin.  (first-latin-draft.pdf)

4. Then we started looking closer at x-height matching. with the proportions we originally worked on Latin would be really small and it already looked really small. (Please see this video for detailed description) So this was the development till last week.
   

All the files will be on GitHub by weekend and will post regularly. We work on FontForge. We are posting weekly meetings here on YouTube and ill post the links on this tread. Its already been a great learning experience and look forward to your comments.

[Pathum Egodawatta]

Hi!

This week Pushpa, myself and two others worked on figuring out and trying to match Sinhala height with Tamil and Abhaya Latin draft, that Sol is working on. We were trying to find solutions for few problems Sol raised in our last call, about x height matching.

We’d like to hear your comments on few specific things. Please note that the latin is being scaled to test the height on this test doc.

1. Comments on matching Sinhala with Latin.

2. Comments on Tamil redesign to match Abhaya Sinhala.

3. Comments on Abhaya Tamil design.

4. Anything else on the project :)

Thanks! The files and notes are posted on GitHub as well.

Cheers,

Pathum

[Pathum Egodawatta]

Hi All,

Please see the latest test docs here.

We are now working on the weight of the Extra Bold. With the high contrast style there is a limit to how far we can increase the Extra Bold weight. Because of this we are having a problem where the difference between the interpolated instances is not significant enough. This week was about this. Please see the latest video call for more in detail description.

https://www.youtube.com/watch?v=9AcZiox-PBk

Cheers,

Pathum

[Pathum Egodawatta]

Hi All,

This week we are still working on figuring out the weights for different versions. Finally we have arrived to a consensus on this issue. As documented on video calls Sol pointed out that with the Sinhala ExBold weight abhaya might lack contrast between the instances. But Abhaya Sinhala is already really Heavy and if we try to make the ExBold heavier we might face an issue of style. Original designer Pushpanada was persistent about maintaining the style and for me it was interesting to see this collaboration effort.

Please see the latest docs posted here and here.

Cheers,

Pathum

[Dave Crossland]

On 29 June 2015 at 16:12, Pathum Egodawatta <path...@gmail.com> wrote:

Please see the latest docs posted here and here.

I recommend shifting to HTML based test docs, and switching you main branch of development to the gh-pages branch, since these fonts are primarily intended for use on mobile web text, so you can easily test each time you push to the repo :) 

[Pathum Egodawatta]

Yeah, actually that makes sense. Will do that tomorrow. 

BTW since Sparkleshare only pushes to Master or the available default branch, when few people are working have to work has to  be on branches or forks.

PS: I forked your Anthoza.py to just generate the OTF and HTML. Have to find a way to filter out the Glyphs only in the font. 

--
You received this message because you are subscribed to the Google Groups "Google Fonts Discussions" group.
To unsubscribe from this group and stop receiving emails from it, send an email to googlefonts-dis...@googlegroups.com.
To post to this group, send email to googlefon...@googlegroups.com.
Visit this group at http://groups.google.com/group/googlefonts-discuss.
To view this discussion on the web visit https://groups.google.com/d/msgid/googlefonts-discuss/CAEozd0wkqzSrNS4COhdtru8mv-ANd2ygqaNxSwEYfBkuK7x-Vg%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.

[Dave Crossland]

On 29 June 2015 at 17:08, Pathum Egodawatta <pat...@mooniak.com> wrote:

Yeah, actually that makes sense. Will do that tomorrow. 

I had the idea that one can develop the project's sources in the master branch and have Travis CI run the font build process and push the sources plus binaries into gh-pages. Info to do this is in https://github.com/googlefonts/fontbakery/#on-travis

 

BTW since Sparkleshare only pushes to Master or the available default branch, when few people are working have to work has to  be on branches or forks.

Right; I'm actually not so hot on SparkleShare any more, it seems making regular git commits with useful commit messages is better; the python scripts Jeremie made seem like a good idea, and doing something similar with a filesystem observer (like https://pypi.python.org/pypi/watchdog) to watch a working directory and make a set of commits on save, so you dont have to think about it, and then rebasing before you push, could be a nice tradeoff between not having to think about making the commits as you work, just hit Ctrl-S like normal, versus being more control about the push/pull process.

 

PS: I forked your Anthoza.py to just generate the OTF and HTML. Have to find a way to filter out the Glyphs only in the font. 

You can use Adobe Blank in the CSS font-family stack so that characters not in the font don't appear in the page visually even though they are in the DOM.

--

Cheers
Dave

[Pathum Egodawatta]

Hi All,

Today

1. We managed to finalise the weight issue :)  Sol wrote off list 'The extraBold seems to be on the limit with de stroke contrast. This issue works totally different comparing to latin.

May be in latin we are used to see extra bold or black weights with thin strokes and hight contrast.'

2. Set up the repo automation, and testing docs for Pushpananda.

Tomorrow

1. Sol will push the latin files to Repo.

2. Pushpa will work on rest of the ExtraBold glyphs

Questions

1.Once shipped In Google Fonts how is this family going to be organised?   As one family : Abhaya Latin + Sinhala + Tamil or  Two families: Abhaya Latin + Sinhala  and    Abhaya Latin  + Tamil? There was a confusion we faced when discussing about matching scripts with Pria and Sol.

Cheers,

Pathum

On Thursday, June 11, 2015 at 8:50:42 PM UTC+5:30, Pathum Egodawatta wrote:

[Dave Crossland]

On 30 June 2015 at 16:15, Pathum Egodawatta <path...@gmail.com> wrote:

Two families: Abhaya Latin + Sinhala  and    Abhaya Latin  + Tamil?

Yes

[Erin McLaughlin]

Hi Pathum! 

I'm compiling a big "GWF Project Reviews" document so I can keep track of when people request reviews. I have your new Github repo address, but I am wondering if you can point me to exactly where I should be looking for your most current test documents?

Thanks,

Erin

[Pathum Egodawatta]

Hi Erin!

Lovely! 

You can see the pdf samples here. https://github.com/mooniak/abhaya-libre-font/tree/gh-pages/documentation/development

Ill  post weekly pdfs here.

And the latest work will be updates here regularly from next week onwards. http://mooniak.com/abhaya-libre-font/tests/

Cheers,

Pathum 

[Pathum Egodawatta]

Today,

1. Woo, We are done with the basic character set. See the repo. 

2.  I showed how Mr Pushpananda to post here and get comments.

3. And OpenType is done as I posted on Ayanna Thread

Tomorrow

1. Make the test docs for review.

2. Pushpa will post here the latest updates.

On Thursday, June 11, 2015 at 8:50:42 PM UTC+5:30, Pathum Egodawatta wrote:

[Pathum Egodawatta]

Hi All,

Today 

- I Updated the Test page and now it will have the latest work fresh from sources.

- Please see http://mooniak.com/abhaya-libre-font/tests/ for your feedback.

Tomorrow

- Pushpa will continue on developing the consonant+matra forms.

Wrapping up early today.

On Thursday, June 11, 2015 at 8:50:42 PM UTC+5:30, Pathum Egodawatta wrote:

[Erin McLaughlin]

Hi Pathum!

I've been struggling to write a review for this project because I really don't know what kind of improvements to suggest; the curves are drawn extremely well, and the forms themselves, and the areas of darkness all follow the conventions of the sort of "classic" Sinhala text faces.

One thing I can suggest at this point, is putting some of the latin letters in the same paragraph as the Sinhala. Or, make a few numerals in this early stage, and put those in the text (like "2015" for example, since that sort of Latin/Sinhala usage is more common). Then we can better assess if the Latin is properly scaled to the Sinhala. I think it's worth to test this before you get too far. Based on my quick photoshop test, I have a feeling that the numbers might seem a bit large?

Sorry that this is not super helpful :)  But it is looking very good so far, to me! Thanks for posting!

-Erin

[Pathum Egodawatta]

Hi Erin!

Thank you! Your concern is valid, I will update the tests shortly and let you know. Yes, infact Abhhaya Sinhala is based off of old letterpress blocks from Chitra Letter Manufacturers in Sri Lanka.

I saw somethings that could be considered to revisit in the design. What do think?

This  matra VolalicR and Rr looks a bit out of place with the mono-linear  loop?

The 'Sannyaka Loop' on red marked glyphs could be a bit more consistant. And on the first glyph the terminal of the sannyaka loop is almost touching the pointer of the base form a bit disturbing, and thoughts?

I will discuss this with Pushpa as well, but I would like to know what you think as well.

Cheers,

Pathum

[Pathum Egodawatta]

Hi All,

Today,

1. More Updates to Sinhala Regular- Basic glyphs alomost complete

Tomorrow

1. Make test docs with latest latin.

On Thursday, June 11, 2015 at 8:50:42 PM UTC+5:30, Pathum Egodawatta wrote:

[Pathum Egodawatta]

Hi All,

Today

1. Made the latest test docs. Available online here: http://mooniak.com/abhaya-libre-font/tests/  Please see this  pdf sample too.

Inde

Tomorrow

1. Call with Sol and Pushpa to decide on weight issue.

2. Fix the weight issues.

3. Generate all the instances and update sample to have all the weights.

Questions/Notes

Seems like we have a weight problem on the regular. And as Erin suspected the Numerals have a height problem too. We will work on this tomorrow.

On Thursday, June 11, 2015 at 8:50:42 PM UTC+5:30, Pathum Egodawatta wrote:

[Pathum Egodawatta]

Hi All,

I asked Rafael Saraiva for feedback on Abhaya Libre. He designend Serendip Sinhala+Latin for his MATD project.  Thanks Rafael!

Off the list he wrote:

_____________________________

Sinhala looks great, the text texture is beautiful. But I have the feeling that Latin looks dark to my eyes when compared to Sinhala. I understand that the general relationship Sinhala-Latin in regard to proportions might be the best considering what you guys consider the standard in Sri Lanka (my non-trained eyes tend to think that Latin is big). But still, I think both scripts do not share the same colour. In addition, the figures look big, even comparing to the Latin only [and the spine of the /eight is quite heavy, no?].

Here I leave my comments on the Sinhala letters [Sorry, I didn't manage to illustrate it]

-- When the heavy strokes sit on the top of the letters as in /i/ka/ta/na/ba/, I have the impression that it looks heavier than the heavy strokes at the baseline. 

-- Perhaps the bottom heavy stroke in /na/ is a tad heavy?

-- I was under the impression that overall /ayanna is a tad heavy. 

-- the virama form in /ee/ is a touch high, no?

-- I wonder why the in-stroke in /nda/ is different from the other letters from the same group.

-- Not sure about the solution on the loop of /chha/. It leaves a tiny white spot on the left that could disappear in small sizes.

-- could the loop of /ja/ have a corner inside? Looks like a circle was inserted there. In display sizes it could stand out. 

-- The exit strokes have different solutions. Sometimes the thin stroke ends perpendicular to the outline, sometimes it has a 90 degree cut either on vertical or horizontal direction. Again, in small sizes it wouldn't be noticeable at all, but since you are planning the extreme weights, the terminals could stand out in bigger sizes.

_____________________________________

[Pathum Egodawatta]

On Sunday, July 12, 2015 at 12:45:37 AM UTC+5:30, Erin McLaughlin wrote:

Hi Pathum!

I've been struggling to write a review for this project because I really don't know what kind of improvements to suggest; the curves are drawn extremely well, and the forms themselves, and the areas of darkness all follow the conventions of the sort of "classic" Sinhala text faces.

One thing I can suggest at this point, is putting some of the latin letters in the same paragraph as the Sinhala. Or, make a few numerals in this early stage, and put those in the text (like "2015" for example, since that sort of Latin/Sinhala usage is more common). Then we can better assess if the Latin is properly scaled to the Sinhala. I think it's worth to test this before you get too far. Based on my quick photoshop test, I have a feeling that the numbers might seem a bit large?

Hi Erin, Please see the latest testing docs. 

http://mooniak.com/abhaya-libre-font/tests/

[Erin McLaughlin]

Hi Pathum! Thanks for sending the link and for the reminder!

I read Rafael's comments that you posted above - I agree with all of those! I tried to come up with a few more suggestions, and am attaching a .pdf with visuals. I also messed around in photoshop, trying to scale the Latin and numerals down to a size that might work a bit better for the Sinhala.

Thank you so much for your patience, and organization! :)

-E

[Pathum Egodawatta]

Hi All,

Today

1. Sol worked on Latin weights and Numerals. SO you can see the updated version live here.

2. Pushpa did fixed some of the problems that Erin and Rafael pointed out in the Sinhala ExtraBold. I will post detailed explanation on changes tomorrow.

Tomorrow

1. More updates to Sinhala

Note

- Abhaya Latin development will pause for next few weeks till we get our hands on Modular Tamil fonts that will become Abhaya Libre Tamil. Team will focus on Ayanna Narrow. But Sinhala development will go on

On Thursday, June 11, 2015 at 8:50:42 PM UTC+5:30, Pathum Egodawatta wrote:

[Pathum Egodawatta]

[Pathum Egodawatta]

[Pathum Egodawatta]

Hi All, 

We have posted the latest updates with fixes to some of the feedback posted above on our GitHub.

[Pathum Egodawatta]

Hi Erin and All!

We Sol updated the Latin with new numerals and weight fixes to Regular. Pushpa updated the Sinhala with some fixes, but about the 'Nga' (U+0d9e), the correct form of this is that shape comes out. In original FM Abhaya it was grouped in /Va /Da group, as you suggested. But Pushpa has been asked to do this change by few linguists. And in some of the Primers which used Abhaya they drew this glyph separately. I have to run this by Sumantri to get confirmation. I will post here when I have more info.

Pushpa and myself sat down to go through the comments. Here is the video. Its in Sinhala, posting for archival purpose :) 

[Pushpananda Ekanayaka]

Today-

1. I did some improvements and adjustments for /i/ka/ta/na upper heavy strokes.

2. Adjusted the bottom heavy stroke in /na/

3. Height adjustment for /ee/

4. change the in stroke in /nda/ like others.

5. Adjusted the loop of /chha/

6. Adjusted the inner loop of /ja/

7. equalised the exit stroke of /ka/ta/na/ga

On Thursday, June 11, 2015 at 8:50:42 PM UTC+5:30, Pathum Egodawatta wrote:

[Dave Crossland]

Hi Sir

Thank you for the daily update - I'm sorry it is delayed in appearing in the group, the first time anyone posts then the message is held for me to approve (to avoid spam) and I have been offline at the http://2015.granshan.com conference the last days. 

I would like to suggest 3 parts to an update. (1) what you did (2) what you'll do next (3, optional) any questions :) 

Cheers,
Dave

--

You received this message because you are subscribed to the Google Groups "Google Fonts Discussions" group.
To unsubscribe from this group and stop receiving emails from it, send an email to googlefonts-dis...@googlegroups.com.
To post to this group, send email to googlefon...@googlegroups.com.
Visit this group at http://groups.google.com/group/googlefonts-discuss.

To view this discussion on the web visit https://groups.google.com/d/msgid/googlefonts-discuss/de929f93-b018-471b-ac13-9843697f4d66%40googlegroups.com.

[Dave Crossland]

When do you think this will wrap up?

[Pathum Egodawatta]

Hi Dave!

We can wrap up Sinhala in two weeks. Latin is on hold till we get Tamil, but we can push and finish Latin for a Sinhala+Latin delivery.  And then work on Tamil+ Latin.

But since there are no overlapping Unicode in Tamil and Sinhala it will ship together?

Cheers,
Pathum

On Thu, Jul 30, 2015, 7:02 AM Dave Crossland <da...@lab6.com> wrote:

When do you think this will wrap up?

--

You received this message because you are subscribed to the Google Groups "Google Fonts Discussions" group.
To unsubscribe from this group and stop receiving emails from it, send an email to googlefonts-dis...@googlegroups.com.
To post to this group, send email to googlefon...@googlegroups.com.
Visit this group at http://groups.google.com/group/googlefonts-discuss.

To view this discussion on the web visit https://groups.google.com/d/msgid/googlefonts-discuss/CAEozd0y0aFCKCqRsmadzrOE3OGsFJM8uRSp8B%3DJVWpLr8sxn1A%40mail.gmail.com.

[Pushpananda Ekanayaka]

Hi All,

1. Last week I finished Abhaya Libre Base characters.

2. Today finished 100 modifiers.

3. Next 3 days I can finished all modifiers.

4. Next week I have to complete base characters of Extra Bold.

[Pathum Egodawatta]

And, I

Today

1. Fixed the feature errors and added feature file to masters. This is for testing purposes only, we will use FDK for building final versions.

Now you can see the features working on the bottom of the testing page.

2. Timeline workout on Tamil 

Tomorrow

1. I don't have much to do in Abhaya project at the moment. Mr Pushpanada will continue updating from here onwards.

[Pathum Egodawatta]

Today and last two days,

1. Tamil properly narrow now. I wa sstruggling with /Nna and Na. Akhand tamil was a good example on Narrow tamil.  :)

2. Build system is in place, I will publish the files to a github tomorrow

3. Met with a documentary maker who is going to make a documentary on Fonts in Sri Lanka

Tomorrow

1. Publish Build System

2. More work in Tamil

3. Call with Sol

[Pathum Egodawatta]

Hi All,

We are in the process of interpolation debugging and fixing small issues. And testing our Open type features on different platforms and different cases. THis will continue for next two days I expect. And then we will work on adding any extra glyphs that we thing nice to have.

Cheers,

Pathum

[Pathum Egodawatta]

You can see the latest test page updated with all weights here http://mooniak.com/abhaya-libre-font/tests/

[Pathum Egodawatta]

Hi All,

Sinhala is complete! you can see the live testing page here http://mooniak.com/abhaya-libre-font/tests/

There are bits and pieces of features that need more work. But so far so good.

Next two days,

1. Test on multiple platforms

2. Get final Latin from Sol and combine it to the FDK build

3. Work with Liang to smooth out the build process.

[Liang Hai]

@Pathum: I've set up the build directory properly and sent a pull request: https://github.com/mooniak/abhaya-libre-font/pull/9

[Pathum Egodawatta]

Hi Liang and all!

Thanks! Im setting up and let you know how it goes. We have some updates to the features as well.

And Question: What do you suggest to do about merging Latin files with Sinhala? Building with FDK itself rather than using pyftmerge afterwards is easy and makes sense I think.

Is there a a way to merge two or more ufos with FDK? 

Cheers,

Pathum

[梁海]

Basically FDK only handles the work from UFOs to OTFs. Merging UFOs is not FDK's duty, neither does it know how to do it, afaik.

It's very easy to merge UFOs using RoboFab (and I already utilize RoboFab in Hindkit, so it's not gonna an extra dependency). A short Python script should do. After that the source directory will be more streamlined — build.py merges the Sinhala UFOs (exported from FontForge) and the Latin UFOs (exported from Glyphs) into the UFO masters (which can even be ignored by .gitignore to avoid data duplication) for FDK.

Maybe Ayantha could have a try first? If you guys run into any trouble of merging UFOs, I'll be able to work on it after 1 September.

梁海 Liang Hai

--
You received this message because you are subscribed to a topic in the Google Groups "Google Fonts Discussions" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/googlefonts-discuss/ET5kBjYxDiY/unsubscribe.
To unsubscribe from this group and all its topics, send an email to googlefonts-dis...@googlegroups.com.

To post to this group, send email to googlefon...@googlegroups.com.
Visit this group at http://groups.google.com/group/googlefonts-discuss.

To view this discussion on the web visit https://groups.google.com/d/msgid/googlefonts-discuss/9b9af759-52df-4c6c-a3d2-285c92ec708e%40googlegroups.com.

[Pathum Egodawatta]

Hi Liang!

Thanks! Ayantha will work on it :)

To unsubscribe from this group and all its topics, send an email to googlefonts-discuss+unsub...@googlegroups.com.

[Pathum Egodawatta]

Hi Liang and all,

On Mon, Aug 31, 2015, 10:38 AM Pathum Egodawatta <path...@gmail.com> wrote:

Hi Liang!

Thanks! Ayantha will work on it :)

On Friday, August 28, 2015 at 7:10:18 PM UTC+5:30, Liang Hai wrote:

Basically FDK only handles the work from UFOs to OTFs. Merging UFOs is not FDK's duty, neither does it know how to do it, afaik.

It's very easy to merge UFOs using RoboFab (and I already utilize RoboFab in Hindkit, so it's not gonna an extra dependency). A short Python script should do. After that the source directory will be more streamlined — build.py merges the Sinhala UFOs (exported from FontForge) and the Latin UFOs (exported from Glyphs) into the UFO masters (which can even be ignored by .gitignore to avoid data duplication) for FDK.

We couldn't  figure out a  proper way  to do this. Ayanatha is working on the features at the moment and would be great if you can give it a go. :)

And just a thought, would you consider making it a part of HindKit? To be able to have 2 or more separate UFO's for a single master. And merge them in the build process? In a collaborative project like Abhaya it would be really useful.

To unsubscribe from this group and all its topics, send an email to googlefonts-dis...@googlegroups.com.

To post to this group, send email to googlefon...@googlegroups.com.

Visit this group at http://groups.google.com/group/googlefonts-discuss.
To view this discussion on the web visit https://groups.google.com/d/msgid/googlefonts-discuss/9b9af759-52df-4c6c-a3d2-285c92ec708e%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

--

You received this message because you are subscribed to the Google Groups "Google Fonts Discussions" group.

To unsubscribe from this group and stop receiving emails from it, send an email to googlefonts-dis...@googlegroups.com.

To post to this group, send email to googlefon...@googlegroups.com.
Visit this group at http://groups.google.com/group/googlefonts-discuss.

To view this discussion on the web visit https://groups.google.com/d/msgid/googlefonts-discuss/7ca04534-0361-4844-a15b-e2be5fd8f7a7%40googlegroups.com.

[Dave Crossland]

On 2 September 2015 at 16:31, Pathum Egodawatta <path...@gmail.com> wrote:

We couldn't  figure out a  proper way  to do this.

What did you try?

[Pathum Egodawatta]

Ayantha couldn't find anything that he could start on. Well we didn't put enough time to understand RoboFab :) 

[Liang Hai]

Ah, well, I happened to tried it a bit when I worked on another project.

Turned out it's indeed easy, something like this:

font_target = robofab.world.OpenFont(FONT_TARGET_PATH)
font_source = robofab.world.OpenFont(FONT_SOURCE_PATH)

new_glyph_names = [i for i in PREDEFINED_GLYPH_NAMES if i not in font_target.keys()]
for glyph_name in new_glyph_names:
    glyph = font_source[glyph_name]
    font_target.insertGlyph(glyph)

font_target.save(OUTPUT_PATH)

Above is the main logic I use, since I need to control what glyphs to be inserted, and I don't want anything to be overridden.

However, RoboFab's case-insensitivity is very dangerous (I tried it first on a Tamil family, and ITF's naming scheme on Tamil-style scripts relies on case-sensitivity), therefore I took a workaround: now I use RoboFab's "ufo3k" branch and use Defcon's "ufo3" branch to call it (because "RoboFab ufo3k"'s objects are not complete yet…).

Well it worked, anyway. But I found it might be better to just use an unofficial fork of RoboFab, say, Cosimo Lupo's https://github.com/anthrotype/robofab/tree/fix_case_insensitive_opt

[Pathum Egodawatta]

---------- Forwarded message ---------
From: Ayantha Randika <paara...@gmail.com>
Date: Tue, Sep 8, 2015 at 5:13 PM
Subject: Merger issue
To: Pathum Egodawatta <path...@gmail.com>

I have created a simple script to merge Latin and Sinla .ufo files in to one .ufo object. Script is attached herewith. When compiled output ufo from AFDKO the compiled otf file does not work. In Windows it gives the error "The requested file is not a valid font file".

Then using font tools, I got a ttx dump of original.otf file (original.ttx). Then again I converted ttx file to a otf font. That otf font works without any issues. 

original.otf >> original.ttx >> font_tools.otf

(font_tools.otf works)

Then I got a ttx dump of the tont_tools.otf named as  tont_tools.ttx. 

font_tools.otf >>  tont_tools.ttx

And then I diff the two ttx files, font_tools.ttx and original.ttx. The only difference of two files is at 'checkSumAdjustment' and 'modified value'. (lines 937 and 942).

I can't figure out why original.otf does not work while font_tools.otf works while the only difference of files is 'checkSumAdjustment' and 'modified value'. Can anyone shed some light on this matter?

All files are attached here.

[Zar]

Hi Pathum,

I don't know with which tool you are making the diff test, but I have used ftxdiff and it shows a change in the cmap subtables, an imporntant one.

I run ftxdiff -l ~/Desktop/original.otf ~/font_tools.otf and it seems that the space character is in Glyphs Index 924 in the original font, when according with the OpenType specifications this glyphs must be the first encoded glyphs (next to .notdef): .notdef, space, … That explain why the original font triggers a error message.

##### Tables changed

'cmap'

##### Table 'cmap' (1098 bytes in the modified font)

                   (1112 bytes in the original font)

### Changed cmaps:

Platform: Unicode

Script  : 2.0 or later (no surrogates)

Language: Unspecified language

### Added mappings (in modified font and not original font):

### Deleted mappings (in original font but not in modified font):

### Changed mappings (char code ==> original font glyph ==> modified font glyph):

U+0020 ==> 924 ==> 1

Platform: Microsoft

Script  : Unicode

Language: Unspecified language

### Added mappings (in modified font and not original font):

### Deleted mappings (in original font but not in modified font):

### Changed mappings (char code ==> original font glyph ==> modified font glyph):

U+0020 ==> 924 ==> 1

### End changed cmaps:

[Dave Crossland]

Hi

Also thanks for your sample text for Sinhala script :) Writing original ones based on character frequency analysis is definitely the best I think. 

Sentence:

වජ්‍රාසන ආරූඪ වූ ශ්‍රී ශක්‍ර දිව්‍ය රාජයා විචක්ෂණ ලෙස බැලුම් හෙළන කල දෝලාවෙන් බට මහා රාජිණිය දුටුවේය.

Translation (google translate):

Vajrasana attributable saw Sri sakra raājaiṇaiya envoy for the West hurled balloons potpourri divine once roamed.

(lol)

Translation (hand):

When the 'Shakra' the king of God is looking at the world he saw the great queen getting down from the palanquin.

Paragraph:

සිංහල අකුරු, ඈත අතීතයේ පටන් ඒවා පරිහරණය කළ අපේ මුතුන් මිත්තන් අතින් වැඩි දියුණු වෙමින් අප කරා පැමිණ තිබෙනවා. සිංහල අක්ෂර නිර්මාණය හා විකාශනය දීර්ඝ ඉතිහාසයකට උරුමකම් කියන්නක්. එය, අපේ ඉපැරැණි චිත්‍ර කලා සම්ප්‍රදාය තුළ ඉතා සුන්දර ප්‍රකාශනයක් හැටියට මතු වුණා. සිංහල අකුරුවලට උපතින් ම හිමි වී තිබෙන ජීව ගුණය, ආවේණික හැඩ තල හා සෞන්දර්ය වටිනාකම එදවස කලා ශිල්පීන් මනා ලෙස හඳුනාගෙන සිටි නිසයි ඒ. මුද්‍රණ ශිල්පයේ සම්ප්‍රාප්තියත් සමඟ, තාක්ෂණය හා මුසු වූ අක්ෂර හැඩ තල හා අනුපාත, ප්‍රායෝගික ව වඩාත් නිවැරැදි හා පැහැදිලි තත්ත්වයක් කරා ළඟා වුණා. අප රටේ මුල් කාලයේ මුද්‍රිත පොතපතේ පටන්, එක්දහස් නවසිය අසූ ගණන් දක්වා ම සිංහල අකුරුවල රූපමය ලක්ෂණ වර්ධනය වුණා. එකල කුඩා දරුවන් ගේ පොතකට ගැනුණු ඊයම්වලින් තැනුණු අකුරු මෙන් ම, චිත්‍ර ශිල්පීන් විසින් අවස්ථානුකූල ව අතින් ඇඳි අකුරු වුවත් ඕනෑ ම ළමයෙකුට කියවා ගැනීමේ අපහසුවක් තිබුණේ නැහැ.

Translation (google translate): 

The letters, since they are used to in the past has come far from our ancestors to our improves. English characters created and aired a long history to tell. It is a very beautiful expression as raised in our ancient art tradition. Pennsylvania has won the vibrancy of English birth, unique shape and aesthetic value of luster because the artists were identified as well. Samaparaāpataiyata with printing, technology and character shape and rates, practical, reaching the most accurate and clear conditions. We began early printed books in the country, one thousand nine hundred eighty-count growth to the pictorial features Sinhalese Pennsylvania. Then the children's book characters as small autos lead the finalists, the artists did not have any difficulty in reading a child, but the font underlined by the situational.

Please correct the hand translation and provide one for the paragraph :)

Cheers
Dave

[Dave Crossland]

On 12 September 2015 at 23:10, Zar <zar.ni...@gmail.com> wrote:

I run ftxdiff -l ~/Desktop/original.otf ~/font_tools.otf and it seems that the space character is in Glyphs Index 924 in the original font, when according with the OpenType specifications this glyphs must be the first encoded glyphs (next to .notdef): .notdef, space, … That explain why the original font trigger a error message.

FontForge says, The glyph named uni0020 is mapped to U+F033A. But its name indicates it should be mapped to U+0020. I didn't investigate this, but this correlates with your discovery. 

I reported this at https://github.com/mooniak/abhaya-libre-font/issues/20

[Ayantha Randika]

Hi,

Thanks Zar and Dave.

I could correct the error by following your leads. The issue was, there was a glyph in one of the original fonts named as "uni0020". This has happened due to an error happened in design phase. So in the final output there were glyphs for both "space" and "uni0020". I could correct it by correcting the names in UFO.

And thanks again for your help.

Thanks,

Randika

[Pathum Egodawatta]

Hi All,

So after few months of work we have finished the Abhaya Libre v.1.x. We did three releases: 1.000 as a pre release for final testing and then with feedback from Dave, we did the latest release v1.030.

It was really fun and nice experience, thanks to everybody who helped and commented. Im working on a draft small writeup summarising the process and what we learned :) WIll post it in coming weeks. 

We already have few issues listed on Github Issues that will be fixed in the next release.

Question:

Abhaya Tamil which will be a refactored Modular font is in the development. Since there is overlapping unicode, we wont be able to ship it as single family on Google Fonts. So what is the best way to approach this? 

1. Should we have two separate repos for this?

2. Or treat it as a one family in production(in same repo, under same family name) and only consider it as two families for Google Fonts shipping? 

3. Can we keep the same family name for both fonts?

[Dave Crossland]

On 25 September 2015 at 16:20, Pathum Egodawatta <path...@gmail.com> wrote:

1. Should we have two separate repos for this?

2. Or treat it as a one family in production(in same repo, under same family name) and only consider it as two families for Google Fonts shipping?  

3. Can we keep the same family name for both fonts?

In the Fonts API, you can't have the same name for both families... perhaps use Abhaya Libre for the Latin + Sinhala and Something Libre for the Latin + Tamil? Or Abhaya Libre for Latin + Sinhala and Abhaya Libre Something or Abhaya Something Libre for the Latin + Tamil?

Do the overlapping unicodes need to be different? If not, I suggest that (2) can be better - you develop a single repo and build combined font binaries, and use pyftsubset and fontTools to split and rename them into latin-plus-another-script pieces.

[Pathum Egodawatta]

I think I made a mistake, Actually zwj zwnj is the only things they share, but they are not in the range of either script. SO this issue is not actually an issue for Sinhala and Tamil?

Another question: Can you direct me to why we can not ship overlapping unicode families in GF? @dave you mentioned elsewhere it might get fixed by W3C.

[Dave Crossland]

On 25 September 2015 at 17:04, Pathum Egodawatta <path...@gmail.com> wrote:

Another question: Can you direct me to why we can not ship overlapping unicode families in GF? @dave you mentioned elsewhere it might get fixed by W3C.

Its to do with the way unicode-range work.  

[Pathum Egodawatta]

Hi All,

We released Abhaya initially few months ago and already have few releases with fixes. There are some exciting issues filed on GitHub by users asking for fixes if anybody is interested.

https://github.com/mooniak/abhaya-libre-font/issues/39

[Pathum Egodawatta]

Hello people!

After many years we are preparing to release a new version of Abhaya with many updates. Most notable addition is the Tamil design based on the Modular Infotech Shree-Tam-3960 / Shree-Tam-3961 fonts. Original project was built completely using Fontforge based build and we have since migrated the project to Glyphs. Back then GF couldn't ship multiple scripts in one family will need to test and see regressions and figure out how we are going to ship Tamil in the project. 

Latest development is happening here: https://github.com/mooniak/abhaya-libre-font/tree/v2-dev

Cheers,

_pe

[Dave Crossland]

Awesome!!

--

You received this message because you are subscribed to the Google Groups "Google Fonts Discussions" group.
To unsubscribe from this group and stop receiving emails from it, send an email to googlefonts-dis...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/googlefonts-discuss/29faa5f5-01d7-4b2a-848b-ea9f2e213f10n%40googlegroups.com.

[Pathum Egodawatta]

Our friend and type designer Aadarsh Rajan was kind enough to spend some time to do a review of the current Tamil design. His main feedback point was to loosen the space and add a separate wordspace for the Tamil characters. Also had some notes. Posting here for documentation. 

[Dave Crossland]

Great improvements :)