RESOLVED CONFIGURATION CHANGED190290
Some combinations of Hebrew letters and vowel points display incorrectly
https://bugs.webkit.org/show_bug.cgi?id=190290
Summary Some combinations of Hebrew letters and vowel points display incorrectly
Dan Stephenson
Reported 2018-10-04 13:44:26 PDT
Created attachment 351622 [details] PDF with screenshots (from Safari and Chrome) of each category of incorrect Hebrew display Certain combinations of Hebrew letters, vowel points, and punctuation marks are not rendering correctly on some versions of Safari/Webkit. I can verify that I see this problem on Mac with High Sierra Version 10.13.5 and Safari Version 11.1.1 (13605.2.8). I also see this problem on iPhone, Version 11.4.1 (15G77). But I do not see the problem on iPad, Version 10.3.3 (16G60). Specifics of the problem: There are three combinations of Hebrew characters that fail to display correctly in newer versions of Webkit, but display correctly in older Webkit, and in Chrome or Firefox. These are: 1. Shin with dagesh (Unicode x05e9 + x05c1 + x05bc), followed by any vowel point or punctuation mark. This combination occurs 4747 times in the Hebrew Bible. The vowel point appears as a slight bulge on the left arm of the shin character. As an example see הַשָּׁמַ֖יִם, the fifth word in Genesis 1:1. 2. Sin with dagesh (Unicode x05e9 + x05c2 + x05bc), followed by any vowel point or punctuation mark. This combination occurs 1089 times in the Hebrew Bible. The vowel point appears as a slight bulge on the left arm of the sin character. As an example see הַשָּׂדֶ֖ה, the ninth word in Genesis 2:5. 3. Yod plus dagesh plus hiriq [Unicode x05d9 + x05bc + x05b4]. This combination occurs 4219 times in the Hebrew Bible. The hiriq is displayed above instead of below the baseline. As an example see וַיִּקְרָ֨א, the first word in Genesis 1:5. This may also interfere with subsequent vowel points or cantillation marks (atnah [Unicode x0591], occurs 68 times, or qamats [Unicode x05b8], occurs one time at 2 Samuel 3:2).
Attachments
PDF with screenshots (from Safari and Chrome) of each category of incorrect Hebrew display (80.61 KB, application/pdf)
2018-10-04 13:44 PDT, Dan Stephenson
no flags
zip file with simple html file and screenshot images illustrating the problematic display (43.73 KB, application/zip)
2018-10-04 13:46 PDT, Dan Stephenson
no flags
Test (2.25 KB, text/html)
2018-10-05 12:26 PDT, Myles C. Maxfield
no flags
Test case in Mojave Safari (309.77 KB, image/png)
2018-10-05 12:26 PDT, Myles C. Maxfield
no flags
Test case in Mojave Firefox (291.98 KB, image/png)
2018-10-05 12:26 PDT, Myles C. Maxfield
no flags
Dan Stephenson
Comment 1 2018-10-04 13:46:59 PDT
Created attachment 351623 [details] zip file with simple html file and screenshot images illustrating the problematic display
Radar WebKit Bug Importer
Comment 2 2018-10-04 20:41:16 PDT
Myles C. Maxfield
Comment 3 2018-10-05 12:26:10 PDT
Created attachment 351689 [details] Test This is a fantastic bug report, thank you so much! I'm trying to reproduce this bug to determine what is going wrong. This looks like a font-specific problem, so I downloaded Ezra SIL from https://scripts.sil.org/cms/scripts/page.php?item_id=SILHebrUnic2#3fee5fec and I put together the attached page which shows some of the preinstalled fonts on macOS. The HTML file in your zip file says that these tests were done with at least six other Hebrew fonts. Which fonts were those six? I'm having trouble reproducing the problems shown in the screenshots you provided. If you open the attached HTML file, does it look like the attached screenshots? I'm seeing that the behavior for shaping this text has changed between High Sierra and Mojave. Do you have a Mojave machine to test with?
Myles C. Maxfield
Comment 4 2018-10-05 12:26:30 PDT
Created attachment 351690 [details] Test case in Mojave Safari
Myles C. Maxfield
Comment 5 2018-10-05 12:26:48 PDT
Created attachment 351691 [details] Test case in Mojave Firefox
Dan Stephenson
Comment 6 2018-10-05 13:15:20 PDT
Thank you for your prompt response! Some of my font tests were done some weeks ago before I had a good handle on the issue, so I don't remember the whole list of fonts I have seen this error with. I was able to verify today the issue using a couple of "scribe" type Hebrew fonts, KeterAramTsova.ttf and TaameyFrankCLM-Medium.ttf. Using Web Inspector I verified that my "normal view" is using Ezra SIL (as EzraSIL.otf, I believe; I also have SILEOT.ttf and SILEOTSR.ttf on my Mac). For the EzraSIL examples in your two screenshot attachments from Mojave: a) for the יִּ (yod-dagesh-hiriq) sample, the screenshot is different from what I see on High Sierra. It looks correct. b) for the שּׁ and שּׂ (shin/sin-dagesh) samples, the problem is with the next vowel point attached to the letter. For example: שָּׁ One of our developers has a machine with Mojave. We just brought up our Genesis 1 file on that device. Apparently whatever you did in Mojave has fixed the Hebrew display issue. So we can advise our Mac clients with the display issue to update to Mojave. That is great news. As my initial report mentioned, we are not seeing the problem on the current iOS on the iPad, but it is an issue on the current iOS on the iPhone. Are you able to give any information on when your fix will be available on all platforms? Thanks again
Myles C. Maxfield
Comment 7 2018-10-05 13:25:32 PDT
(In reply to Dan Stephenson from comment #6) > Thank you for your prompt response! > > Some of my font tests were done some weeks ago before I had a good handle on > the issue, so I don't remember the whole list of fonts I have seen this > error with. I was able to verify today the issue using a couple of "scribe" > type Hebrew fonts, KeterAramTsova.ttf and TaameyFrankCLM-Medium.ttf. Using > Web Inspector I verified that my "normal view" is using Ezra SIL (as > EzraSIL.otf, I believe; I also have SILEOT.ttf and SILEOTSR.ttf on my Mac). > > For the EzraSIL examples in your two screenshot attachments from Mojave: > a) for the יִּ (yod-dagesh-hiriq) sample, the > screenshot is different from what I see on High Sierra. It looks correct. > b) for the שּׁ and שּׂ > (shin/sin-dagesh) samples, the problem is with the next vowel point attached > to the letter. For example: שָּׁ Ah, I see now. Thanks for the clarification. > > One of our developers has a machine with Mojave. We just brought up our > Genesis 1 file on that device. Apparently whatever you did in Mojave has > fixed the Hebrew display issue. Great to hear it! > > So we can advise our Mac clients with the display issue to update to Mojave. > That is great news. > > As my initial report mentioned, we are not seeing the problem on the current > iOS on the iPad, but it is an issue on the current iOS on the iPhone. Are > you able to give any information on when your fix will be available on all > platforms? Are you seeing the issue on iOS 12? If it works in Mojave, it should work in iOS 12. > > Thanks again
Dan Stephenson
Comment 8 2018-10-05 14:14:26 PDT
You are correct. Our developer pointed out to me that my phone was not in fact on the latest version of iOS. I updated to iOS 12 and checked the test cases. All looks good. I have passed on to our customer support team the information that clients can upgrade to Mojave or iOS 12 to resolve their font display issues. Thanks again for your quick and accurate help.
Note You need to log in before you can comment on or make changes to this bug.