220237 - Unknown character entities display as long whitespace area, can mess up layout on that line

Status: RESOLVED WORKSFORME

(Core Graveyard :: GFX: OS/2, defect)

Description

[Sam Little]

User-Agent:       Mozilla/5.0 (OS/2; U; Warp 4.5; en-US; rv:1.5b) Gecko/20030825
Build Identifier: Mozilla/5.0 (OS/2; U; Warp 4.5; en-US; rv:1.5b) Gecko/20030825

When an unknown or undisplayable character entity (e.g., ™ for TM symbol)
is found in a web page, whitespace for the widest character in the character set
is painted instead. If multiple character entities are found in the same line of
displayed text, positioning of other entities seems to be thrown off.


Reproducible: Always

Steps to Reproduce:
1. Load above page (may be changed as I've complained to them)
2. Look specifically at first line of first paragraph below "What is Republican
Forum and what is a Forum?).


Actual Results:  
I see a block of space following "RepublicanForum.com" where the Trademark
symbol should be. Also, the Registered Trademark symbol that should appear after
"CompuServe" instead appears between the 'r' and the 'v' in "CompuServe" and the
following sentence runs together rather than with the space indicated in the
HTML source:

[...] hosted by CompuServe<sup>&reg;</sup>. The Republican Forum on
CompuServe<sup>&reg;</sup> [...]


Expected Results:  
I don't know what *should* appear for an unresolved character entity, but it
shouldn't throw off other entities and sometimes margins. Pages with
"smartquotes" are particularly troublesome.

Ideally, if the font chosen supports a particular character (for instance, I use
Unicode fonts as defaults), I'd like to see it shown....

This should be OS/2-specific, or at least it isn't a problem in Windows versions....

Comment 1

[Boris Zbarsky [:bzbarsky]]

CTL refers to particular scripts like Thai in which glyph shapes depend on
surrounding glyphs, which is not the problem here....

Comment 2

[jhp (no longer active)]

Unfortunately, there's not much we can do on OS/2 for this.  The system does not
provide the facility to ask a particular font if it supports a given glyph.  So
what we do on OS/2 is draw glyphs outside the 'normal' range using the specified
Unicode font, which for OS/2 is Times New Roman MT 30.  If you don't have that,
then there's not much we can do.

I have been investigating using the font library from Innotek to query whether a
font can render a glyph.  I think there is already a defect for this, but I'll
just keep this one open until I find where to dup it.

Comment 3

[Sam Little]

Okay, what confuses me is that at least some of the problem seems to have been
resolved with 1.5rc1. The page listed now displays correctly, as do at least
*some* pages that use "smart quotes" (the bane of my web-viewing pleasure). 

A test page I played with using "™", "™" and "™" *all* work
now, as do other Unicode entities outside the base character set. It's a *vast*
improvement.

So I guess this is now WORKSFORME. Thanks!

Comment 4

[jhp (no longer active)]

Resolving bug per user's comment