Page MenuHomePhabricator

Conjunct character displayed wrongly on Marathi Wikipedia on Windows; works correctly on Hindi Wikipedia
Closed, InvalidPublicBUG REPORT

Assigned To
None
Authored By
shantanuo
Mar 15 2022, 3:03 AM
Project Tags
Referenced Files
F35013438: image.png
Mar 21 2022, 4:56 AM
F35006575: mr_kalyan_wrong.png
Mar 15 2022, 8:33 AM
F35006542: hindi correct display
Mar 15 2022, 8:33 AM
F35006498: Screenshot from 2022-03-15 09-16-13.png
Mar 15 2022, 8:16 AM

Description

some conjuncts starting with the devanagari character "ल" are displayed incorrectly on Marathi wikipedia site.
It looks "ल्‌य" (with halant) instead of "ल्य" with half ल्‍
For e.g. search for the word "कल्याण" on Marathi (mr) and Hindi (hi) wikipedia and you will see difference.
The hindi wikipedia can display the same word correctly even if it is using the same mediawiki software.
What changes Marathi wikipedia should follow so that it will display the conjuncts correctly?

Event Timeline

Aklapper changed the task status from Open to Stalled.Mar 15 2022, 8:16 AM
Aklapper removed a project: MediaWiki-General.

@shantanuo: Thanks for reporting this. Please always use the bug report form (linked from the top of the task creation page) to create a bug report, and fill in all information in the template. Include screenshots, and include browser information, browser version, operating system:

List of steps to reproduce (step by step, including full links if applicable):

What happens?:

What should have happened instead?:

Software version (if not a Wikimedia wiki), browser information, screenshots, other information, etc.:

Screenshot from Firefox 98 on Fedora 35:

Screenshot from 2022-03-15 09-16-13.png (527×485 px, 82 KB)

List of steps to reproduce (step by step, including full links if applicable):

Type or search any conjuncts starting with the devanagari character "ल". For e.g. "कल्याण" on marathi wikipedia.

What happens?:

It looks "ल्‌य" (with halant)

What should have happened instead?:

It should look "ल्य" as on hindi wikipedia with half ल्‍

Software version (if not a Wikimedia wiki), browser information, screenshots, other information, etc.:
Windows only bug. Linux is OK. Chrome and Firefox both browsers are affected.

Screenshot from chrome (or firefox) on Windows
Marathi incorrect display:

mr_kalyan_wrong.png (768×432 px, 36 KB)

Hindi correct display from the same browser:

hindi correct display (768×607 px, 44 KB)

Does this problem still happen if you go to https://mr.wikipedia.org/wiki/कल्याण?safemode=1 ?
See https://www.mediawiki.org/wiki/Help:Locating_broken_scripts (linked from https://www.mediawiki.org/wiki/How_to_report_a_bug ).

Both https://mr.wikipedia.org/wiki/कल्याण and https://hi.wikipedia.org/wiki/कल्याण look the same in Chromium 98 on Linux, and both sites define font-family: 'Linux Libertine','Georgia','Times',serif; as CSS for the <h1>.

The problem is still there even if safe mode is enabled. The problem can be seen on the entire page and is not limited to heading tag.
It makes it very difficult to read the text.

Does this problem also happen to you when you use a different browser than Chrome (e.g. Firefox or Edge)?
Does this problem also happen to someone else who uses Windows and Chrome/Chromium?

Aklapper renamed this task from conjuncts displayed wrongly on marathi wikipedia to Conjunct character displayed wrongly on Marathi Wikipedia on Windows; works correctly on Hindi Wikipedia.Mar 15 2022, 9:28 AM
Aklapper added a project: I18n.

Bug noted on these windows version:

  • Windows 10 Home Single Language
  • Windows server 2019

Bug is not present on these windows version:

  • windows 7 ultimate

I can upgrade to windows 7 ultimate to address this issue.
But I will like to know if the fonts issue can be fixed without spending time and money on upgrade.
This is a minor nuisance and a casual reader may not even notice it, but an astute reader will be annoyed :)

Does this problem also happen to you when you use a different browser than Chrome (e.g. Firefox or Edge)?

Yes. All browsers are affected. It means this is O/S related bug.

@santhosh: Do you know by any chance which system fonts Microsoft Windows is shipping, and if any workaround could be done about this server-side? Thanks!

I can confirm the bug in Windows 10. The default font applied is Nirmala UI. I just checked the same issue in Windows 11 and the rendering is correct there. Screenshot from Windows 11+Chrome 98

image.png (1,574×904 px, 633 KB)

Nirmala UI series of fonts for Indian languages is known for bugs like these, but being a proprietary licensed font, there is not much we can do about it. It is good that they have fixed it in latest versions. Windows 10 ships with Nirmala UI 1.40. Windows 11 has Nirmala UI 1.41

Thanks for the update. But that does not explain why Hindi Wikipedia has no issues even if it is using the same font!

Thanks for the update. But that does not explain why Hindi Wikipedia has no issues even if it is using the same font!

Eventhough a single font is used for languages using Devangari script, with in the font, the ligature formation rules and glyphs are mapped to specific languages. Marathi language uses a subset of Devanagari script which is not strictly same as the subset of Hindi. The mapping of glyphs and lookups to a specific language is done using locl opentype featue.

(This is the explanation from common opentype engineering practice. I am not sure if Nirmala UI did exactly the same or not. I hope this helps.)

Nirmala UI series of fonts for Indian languages is known for bugs like these, but being a proprietary licensed font, there is not much we can do about it.

Thanks for the explanation! Closing as invalid then ("a change that is outside the power of the component's developers").