Noto Urdu numbers inaccurate

[GoogleCodeExporter]

The Urdu numbers being used in the Noto Urdu fonts at noto are inaccurate. I 
downloaded Noto Nastaliq Urdu and Noto Naksh Arabic for Urdu Pakistan. 

The numbers 4, 6 and 7 being used are Farsi instead of Urdu. This is causing 
issues since people mistakenly use Farsi or Hindi numbers for Urdu. 

Original issue reported on code.google.com by toffeeki...@gmail.com on 1 Feb 2015 at 5:31

[GoogleCodeExporter]

Original comment by stua...@google.com on 4 Feb 2015 at 2:29

[GoogleCodeExporter]

Original comment by roozbeh@google.com on 28 Feb 2015 at 12:31

[GoogleCodeExporter]

I can see that in NotoNastaliqUrduDraft, the digit 7 is not in Urdu style, but 
I don't see that for 4 and 6.  Are you sure you are using digits in the 
U+06F0..U+06F9 and NOT U+0660..U+0669?

Original comment by behdad@google.com on 5 Mar 2015 at 8:11

• Changed state: Accepted

[GoogleCodeExporter]

The 7 will be fixed in the upcoming version of the font.

Original comment by behdad@google.com on 5 Mar 2015 at 8:11

[GoogleCodeExporter]

Interestingly, I can reproduce this on Notepad and BabelPad[1] (screenshots 
attached for both), but not on Microsoft Word (screenshot attached) or Firefox.

In the screenshots, the first line uses U+06F0..U+06F9, and the second one uses 
U+0660..U+0669. In Notepad and Babelpad, digits 4, 6, and 7 use Farsi style 
glyphs instead of Urdu style in the first line. In Microsoft Word (and 
Firefox), Urdu style glyphs are displayed. 

(I am using Windows 7, BTW. NotoNastaliqUrduDraft reports its version as 0.70.)

[1] http://www.babelstone.co.uk/Software/BabelPad.html

Original comment by saadat.m...@gmail.com on 11 Mar 2015 at 7:11

Attachments:

• digits-notepad.png
• digits-babelpad.png
• digits-ms-word.png

[GoogleCodeExporter]

After Notepad and BabelPad, I have noticed this issue in Internet Explorer 11 
as well.

Here's a test page: http://files.saadatmand.pk/type/noto-nastaliq-digits.html . 
When no language is specified in HTML markup, Firefox displays the digits (4, 
6) in Urdu style, whereas IE 11 shows them in Farsi style. However, both 
browsers show Urdu style glyphs if Urdu _is_ specified in the markup.

Original comment by saadat.m...@gmail.com on 11 Mar 2015 at 4:29

[GoogleCodeExporter]

Now that's *very* interesting.

My only guess is that parts of Windows, when no language tag is provided, tag 
the U+06F0..9 digits as Persian (Farsi), which will then trigger the 
substitution in the font.  If that is indeed the case, I don't think there's 
much of a way out of this. :(

Original comment by beh...@chromium.org on 17 Mar 2015 at 1:06

[GoogleCodeExporter]

I can in fact reproduce this using the Uniscribe backend in HarfBuzz.

Original comment by beh...@chromium.org on 17 Mar 2015 at 1:11

[GoogleCodeExporter]

[deleted comment]

[GoogleCodeExporter]

The first time I tried it was using the Urdu language on the language bar. Can 
the Farsi numbers simply be removed from the Urdu font? Or some swapping or 
something to ensure its them and not Farsi/English numbers that get picked? 
Otherwise the problem may show itself on multiple devices not just windows.

Original comment by toffeeki...@gmail.com on 17 Mar 2015 at 2:52

[GoogleCodeExporter]

Humm.  Although we are confident this is a Windows-only issue, I think I can be 
convinced about that proposal.  Let me think about it.  Roozbeh, what do you 
think?

Original comment by behdad@google.com on 17 Mar 2015 at 4:01

[GoogleCodeExporter]

It's OK as a temporary patch, but people in Pakistan and India would like to 
set Persian text in the font, I'm sure.

We can just drop them in a pre-release process.

Original comment by roozbeh@google.com on 17 Mar 2015 at 4:07

[GoogleCodeExporter]

There's not much temporariness about it.  We either decide to ignore the 
Uniscribe issue, or the font that reaches users won't have the language 
mapping.  Even if glyphs are in there but not reachable, that wouldn't make 
much difference to the user.

Original comment by behdad@google.com on 17 Mar 2015 at 4:36

[GoogleCodeExporter]

Temporarily until we figure out what Windows is doing, then we can make it 
permanent ;)

Original comment by roozbeh@google.com on 17 Mar 2015 at 5:00