727560 - Hebrew diacritics are not displayed correctly since Firefox 10

Status: RESOLVED WORKSFORME

(Core :: Layout: Text and Fonts, defect)

Description

[Amir Aharoni]

Attachment: niqqud-test.html

User Agent: Mozilla/5.0 (Windows NT 5.1; rv:12.0a2) Gecko/20120215 Firefox/12.0a2
Build ID: 20120215042007

Steps to reproduce:

I opened a simple file with some Hebrew with diacritics (attached).


Actual results:

The diacritics were displayed somewhere to the side from where they are supposed to be.


Expected results:

The diacritics should have been displayed correctly.

Comment 1

[Amir Aharoni]

Attachment: correct and incorrect display of diacritics

A screenshot of LibreOffice and Aurora. LibreOffice is showing the same text in the same font correctly.

Comment 2

[Boris Zbarsky [:bzbarsky]]

Worksforme on Mac...

Comment 3

[Amir Aharoni]

It may be a Windows-only or an XP-only bug.

It's hard for me to check now, but if i'm not mistaken, this started happening in Firefox 10 and the latest builds of Nightly and Aurora. Before that the diacritics were displayed correctly.

Comment 4

[Simon Montagu :smontagu]

Apparently XP-only, at least I can't reproduce on 7.

Comment 5

[Jonathan Kew [:jfkthame]]

WFM on Win7, too, though I notice that I'm seeing a different font than is shown in the screenshot.

What is the font being used on XP? (Use fontinfo add-on...)

Does it make any difference if you set gfx.font_rendering.harfbuzz.scripts to 7 in about:config (instead of the default value of 3)?

Comment 6

[Amir Aharoni]

The font is Arial.

Setting gfx.font_rendering.harfbuzz.scripts to 7 doesn't change the rendering.

Comment 7

[Alice0775 White]

If I do not misread str, regression window is as follows:
And I cannot reproduce in Firefox12beta but reproduce in Aurora13.0a2 and Nightly14.0a1 on Windows XP only.

Regression window(m-c)
Good:
http://hg.mozilla.org/mozilla-central/rev/e61a169463c6
Mozilla/5.0 (Windows NT 5.1; rv:13.0) Gecko/20120222 Firefox/13.0a1 ID:20120222095919
Bad:
http://hg.mozilla.org/mozilla-central/rev/49956e0cc0bb
Mozilla/5.0 (Windows NT 5.1; rv:13.0) Gecko/20120222 Firefox/13.0a1 ID:20120222142417
Pushlog:
http://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=e61a169463c6&tochange=49956e0cc0bb


Regression window(m-c)
Good:
http://hg.mozilla.org/integration/mozilla-inbound/rev/83c85131bcfe
Mozilla/5.0 (Windows NT 5.1; rv:13.0) Gecko/20120222 Firefox/13.0a1 ID:20120222005219
Bad:
http://hg.mozilla.org/integration/mozilla-inbound/rev/063e67b9f910
Mozilla/5.0 (Windows NT 5.1; rv:13.0) Gecko/20120221 Firefox/13.0a1 ID:20120222022318
Pushlog:
http://hg.mozilla.org/integration/mozilla-inbound/pushloghtml?fromchange=83c85131bcfe&tochange=063e67b9f910


Suspected:
063e67b9f910	Jonathan Kew — bug 722139 - Turn on harfbuzz shaper by default for Hebrew on all platforms. r=smontagu

Comment 8

[Simon Montagu :smontagu]

The regression window is a little odd, considering comment 6 (which implies that harfbuzz doesn't have any effect on this) and the fact that the bug was originally reported earlier than the "last good" timestamp.

Comment 9

[Simon Montagu :smontagu]

I suppose it's possible that harfbuzz exposed an already existing bug for some configurations of OS and installed fonts.

Comment 10

[Alice0775 White]

(In reply to Simon Montagu from comment #8)
> The regression window is a little odd, considering comment 6 (which implies
> that harfbuzz doesn't have any effect on this) and the fact that the bug was
> originally reported earlier than the "last good" timestamp.

Umm, In my WindowsXP
gfx.font_rendering.harfbuzz.scripts in Nightli14.0a1
0,1,2,3 : OK.
4,5,6,7 : NG.

Comment 11

[Alon Bar-Lev]

Attachment: genesis.png

Linux Firefox 18.0.1 Gentoo

Worked perfectly for years.

Now diacritics are in offset.

Test is available at[1], use culmus fonts.

[1] http://www.mechon-mamre.org/c/ct/c0101.htm

Comment 12

[Alon Bar-Lev]

Attachment: genesis-ff-10.png

Linux Firefox 10.0.111 Gentoo

Perfect rendering.

Test is available at[1], use culmus fonts.

[1] http://www.mechon-mamre.org/c/ct/c0101.htm

Comment 13

[Alon Bar-Lev]

Amir Aharoni,

Can you please set platform to all or at least add Linux?

Summary:

Firefox-10 - OK.
Firefox-17 - Regression.
Firefox-18 - Regression.

If you think this bug is Windows specific I will open a new bug for Linux.

Thanks!

Comment 14

[Amir Aharoni]

http://www.mechon-mamre.org/c/ct/c0101.htm looks good to me with Culmus fonts on Firefox 18 in Fedora 18 and in Windows XP.

Comment 15

[Alon Bar-Lev]

(In reply to Amir Aharoni from comment #14)
> http://www.mechon-mamre.org/c/ct/c0101.htm looks good to me with Culmus
> fonts on Firefox 18 in Fedora 18 and in Windows XP.

This is strange, can you attach a picture please?
Do you have the original issue reported?

Comment 16

[Alon Bar-Lev]

Attachment: genesis.html

Sample with multiple fonts.

Frank Ruehl CLM, David CLM shows offset.

Liberation Sans, DeJaVu Serif behaves correctly.

Guess: difference between type 1 fonts and true type fonts?

Comment 17

[Jonathan Kew [:jfkthame]]

I tried installing the current version of David CLM (from sourceforge, v 0.120, although the download archive was named culmus-0.130). The genesis.html sample looks pretty good here (viewing on OS X), although I notice one issue: just about every word contains at least one character (one of the Hebrew accent marks) that isn't supported in the David font, and so falls back to some other font on the system.

This means that the results you see will vary depending what particular fallback font is found for those accents on your system. You might find that they look reasonably OK together - or you might not.

It also means that David CLM, at least (and quite likely other fonts from the culmus package) is not really suitable for fully-accented Biblical Hebrew, as it does not support all the necessary characters.

Comment 18

[Alon Bar-Lev]

Thank you for testing this!

(In reply to Jonathan Kew (:jfkthame) from comment #17)
> It also means that David CLM, at least (and quite likely other fonts from
> the culmus package) is not really suitable for fully-accented Biblical
> Hebrew, as it does not support all the necessary characters.

However, how can we explain that all worked well in firefox-10?

Comment 19

[Alon Bar-Lev]

Attachment: genesis-html-ff-10.png

genesis.html as viewed with firefox-10

Comment 20

[Alon Bar-Lev]

Attachment: genesis-html-ff-17.png

genesis.html as viewed with firefox-17

Comment 21

[Simon Montagu :smontagu]

I get better results with Taamey Frank CLM and Taamey David CLM, in fact I only have these versions installed, and not the non-Taamey ones. I don't know if there's any fontconfig magic that will make the system use these fonts when "David" or "Frank Ruehl" is specified

Comment 22

[Alon Bar-Lev]

(In reply to Simon Montagu from comment #21)
> I get better results with Taamey Frank CLM and Taamey David CLM, in fact I
> only have these versions installed, and not the non-Taamey ones. I don't
> know if there's any fontconfig magic that will make the system use these
> fonts when "David" or "Frank Ruehl" is specified

Thank you for the hint, indeed these fonts are working correctly in firefox-10 and firefox-17. Not sure why the others worked in firefox-10. But I don't mind of using these.

Comment 23

[Ryan VanderMeulen [:RyanVM]]

Is there still an issue here with current Firefox releases?

Comment 24

[Amir Aharoni]

Sorry, I don't have easy access to a Windows machine.

I totally trust Simon to test it.

Comment 25

[Ryan VanderMeulen [:RyanVM]]

Simon, is there anything actionable to do here still?

Comment 26

[Simon Montagu :smontagu]

As far as I know this problem no longer manifests.