351697 - Firefox fails to fall back on MV Boli or FreeSerif to represent Unicode characters from the Thaana script.

Status: NEW

(Core :: Internationalization, defect)

Description

[Benjamin Hawkes-Lewis]

User-Agent:       Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.8.1b2) Gecko/20060907 BonEcho/2.0b2
Build Identifier: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.8.1b2) Gecko/20060907 BonEcho/2.0b2

If text is being displayed in a given font, but includes Unicode characters missing from that font, Firefox is supposed to use other fonts which do include the character if they are available. FreeSerif is one of the few fonts to include characters from the Thaanu script. Internet Explorer 6 and 7 Beta 3 and Opera 9.01  all use FreeSerif, when installed, to display Thaanu characters. But Firefox fails to do so, instead displaying a box with the Unicode code inside, as the example test page I've constructed demonstrates.

Reproducible: Always

Steps to Reproduce:
1. Install Firefox 1.5 or Bon Echo on Windows XP.
2. Install FreeFonts from http://download.savannah.gnu.org/releases/freefont/ .
3. Visit the test page provided or another page containing characters from the Thaanu script (use Alifu (U+0787) as a quick test).

Actual Results:  
Firefox failed to display the Thaanu characters using FreeSerif.

Expected Results:  
Firefox should have displayed the Thaanu characters using FreeSerif, when not available in a preferred font.

Comment 1

[Benjamin Hawkes-Lewis]

Sorry, for some reason I kept spelling the name of the script as "Thaanu" instead of the correct "Thaana".

Comment 2

[Benjamin Hawkes-Lewis]

Firefox fails to fall back on MV Boli, which comes with Windows XP, either. It also fails to display characters which are in both MV Boli and FreeSerif, even when both are included in the text's CSS font-family property.

Comment 3

[Benjamin Hawkes-Lewis]

Update to my last remark: it is possible that the blanking of the subset is okay, as they appear to be diacritical marks on consonants indicating vowels, rather than letters in themselves. Can anyone confirm that part of Firefox's behaviour is correct? 

Comment 4

[Justin Kerk]

Yeah, if you look at the Unicode code chart at http://www.unicode.org/charts/PDF/U0780.pdf the characters U+07A6-U+07B0 are combining characters so I wouldn't expect them to display meaningfully in isolation. (Usually combining characters do appear but offset to the left, but since Thaana is a complex script, the fonts' OpenType instructions may not be set up to display anything in this case).

Comment 5

[Justin Kerk]

Also, the code that does this sort of thing is being massively rewritten for Firefox 3 (and isn't going to be touched at all for Firefox 2), so this bug report isn't that useful unless you see the same behavior with a current trunk ("Minefield") build from http://ftp.mozilla.org/pub/mozilla.org/firefox/nightly/latest-trunk/

Comment 6

[Benjamin Hawkes-Lewis]

On my system, at least, it happens with Minefield too.

Comment 7

[Jungshik Shin]

I haven't installed Free Serif or MV Boil, but probably I have a font that covers Thaana and all the characters in your test page gets rendered fine (Win2k, FF 2.0beta2). There isn't much difference between FF 1.5 and FF 2.0beta2 as far as the rendering engine goes. The only possible difference between your system and mine (other than fonts installed and Windows version) is that you may not have enabled complex script support in the control panel. However, that shouldn't make any difference in the rendering of individual characters. 
 

Comment 8

[Simon Montagu :smontagu]

(In reply to comment #0)
> But Firefox fails to do so, instead displaying a box with the Unicode code
> inside, as the example test page I've constructed demonstrates.

On Windows as opposed to Linux, Firefox itself doesn't display boxes with the Unicode for unavaialable characters. Maybe you have another font installed with glyphs like that for all Unicode characters, which Firefox is picking up before MV Boli or FreeSerif? (which would explain why the "workaround" stylesheet solves the problem)

Comment 9

[Benjamin Hawkes-Lewis]

Further to Jungshik Shin's comment above, from the "Regional and Language Options" I reinstalled "files for complex script and right-to-left languages (including Thai)" and also installed "files for East Asian languages". I also added *all* code page conversion tables for non-unicode programs. Upon restart, my test paged worked even for me.

So I think this was primarily a system misconfiguration issue, although I'm still baffled as to why Opera and Internet Explorer worked before but Firefox didn't. Is it because they supplement fall back information provided by the host OS with their own lists of relevant fonts? As Simon Montagu wondered, I do in fact have a fallback font installed - the Unicode BMP Fallback SIL font[*]. But I don't see why Firefox should have picked that up in preference.

I'll leave it up to a more knowledgeable reviewer to decide whether Firefox's behaviour in that instance is a bug or not, because it seems rather suboptimal to me. If it's not, then this report can be closed.

Vaguely off-topic for this report, I wonder whether it would be worth having a page detailing how to maximize Unicode support in the Firefox help file.

[*] http://scripts.sil.org/cms/scripts/page.php?site_id=nrsi&item_id=UnicodeBMPFallbackFont

Comment 10

[BugBot [:suhaib / :marco/ :calixte]]

The bug assignee didn't login in Bugzilla in the last 7 months, so the assignee is being reset.