339486 - Sometimes monospace text changes partially into another font when it is set to UTF 8.

Status: RESOLVED WORKSFORME

(Core :: Graphics, defect)

Description

[Ria Klaassen (not reading all bugmail)]

Steps to reproduce:
- Go to Bug 337451.
- Set encoding to UTF 8 (happens sometimes spontaneously here) and make sure that monospace is set to Courier New (=default).
- Result: a part of the text ("It's not Bias, It's Discrimination") shows up in another font; Arial Narrow here. Possibly it happens in this bug as well.

Normally the font should be displayed as Courier New, but it seems that Firefox reads a small part of the text as Courier and changes it into Arial Narrow. 

I can't reproduce this consistently; even when I copied the profile to another location and tried it again, the bug failed sometimes to show up.
And sometimes (at the moment now) it even happens in a new freshly created profile. The bug even shows up in the source.

This bug may be related some way to Bug 324706. Possibly it is a dupe, but I could not find a bug that covers the problem.

Comment 1

[Ria Klaassen (not reading all bugmail)]

Attachment: screenshot of the problem

Comment 2

[Ria Klaassen (not reading all bugmail)]

Attachment: screenshow View Source

Comment 3

[Ria Klaassen (not reading all bugmail)]

this is a test: It's not Bias, It's Discrimination. 

Comment 4

[Justin Kerk]

*** This bug has been marked as a duplicate of 332649 ***

Comment 5

[Jean-Marc Desperrier]

I don't really agree with Justin about duping bug that bring some interesting sample case, I'd prefer to mark them as blocked by the original bug, once we have enough information about how to repro them and what they consist of exactly.

But I don't know if this is really a dupe.
The reproduction steps are not good enough for me yet, I don't reproduce at home.

The original text is windows-1252 encoding. Interpreting it as UTF8 should yield invalid characters for the windows-1252 codepoints, displayed as such : black diamond with a white question mark in center. This is what I get.

Instead in the bug, they are handled as valid, but undisplayable code points.

Thebes looks for a font that can display them, finds none, but after the failure seems to stay with one of the font it has tried when looking for a font that could display the characters, instead of continuing to use the original font.

That's two issues in fact :
- not displaying the invalid characters with the correct character : because the black diamond character is not available ? Because the caracter is not really marked as invalid, but contains some random content that does not correspond to any existing character ? 
- After looking for an alternative font and failing, not going back to the original font to display the text. I can not reproduce that either when creating another test case where I mix deliberately not exiting chars from the private area chars with ordinay content.

Comment 6

[Ria Klaassen (not reading all bugmail)]

Could be that I copied the quote from xml pretty print in an older Firefox build (= most plausible), but it is also possible that I copied it from his website, or from Notepad2, or that I typed it directly in the box.
Fact is that it only sometimes showed up in a different font although encoding was set to UTF 8 and that I failed to make a testcase.
Either way, it is strange that Firefox cannot display the ' character under some conditions but not always under those same conditions.

Before I filed this bug I also saw the bug that it was duped to but I did not recognize it as a dupe. Because of the above mentioned reasons.

Comment 7

[Justin Kerk]

Jean-Marc raises some good points, there's not much point in looking into this until bug 332649 gets fixed, but there are some issues here that might not automatically be fixed by it. Reopening and setting dependency.

Comment 8

[Stuart Parmenter]

I suspect this was mostly fixed with the checkins from 340590 or at least the behavior may be changed a bit.  Can someone test it some more?

Comment 9

[Ria Klaassen (not reading all bugmail)]

It shows squares and I think that's correct, for that particular font is apparently missing on my system. When I run the page in Firefox 1.5.0.4 and I set it to UTF 8, I see question marks. So if this is problem, it is not a cairo problem.
The Bug 332649 part of this bug seems to be solved. Everything is displayed in Courier New now.

So I'll close this bug but if anyone else still sees an issue in this, please reopen or file another bug. :)