119664 - Several mathml characters have no transliteration

Status: RESOLVED WONTFIX

(Core :: Layout, defect)

Description

[Hixie (not reading bugmail)]

U+2146 is not yet part of the UNICODE standard, although it has been 
accepted by UNICODE. However, MathML uses it for ⅆ, the 
"d" in "dy/dx". Since I have no fonts installed that contain U+2146, I 
am instead gtting "?y/?x", which is not very pretty.

I suggest we add a hard coded transliteration, for now.

Comment 1

[rbs]

There are several other characters like that. Want to propose all in one go?

Comment 2

[Hixie (not reading bugmail)]

Getting the exact list of what those characters are is a really tedious process
(you have to look up the characters in the DTD and then look to see if they are
in the database... I guess it could be scripted), so I don't think I'll have the
time to do it any time soon. The reason I mention ⅆ is that I am
using that on a daily basis. I haven't run across any others yet. If I spot any,
I'll note them here.

I suggest you fix ⅆ and then reassign this bug to me so that I can
get the fuller list -- would that be ok?

Comment 3

[rbs]

Sounds ok. (Now you get an idea of how tasty these MathML characters are :-)

Comment 4

[rbs]

Attachment: fix

Comment 5

[rbs]

Simple fix, seeking r=shanjian, sr=waterson

Comment 6

[Shanjian Li]

Comment on attachment 65004 [details] [diff] [review]
fix

r=shanjian, (This code point is assigned to MathML, but I don't understand why
unciode standard give it a name like "DOUBLE-STRUCK ITALIC SMALL D". Isn't
"DifferentialD" better? )

Comment 7

[Chris Waterson]

Comment on attachment 65004 [details] [diff] [review]
fix

rs=waterson

Comment 8

[rbs]

Checked in. Now passing the bug to hixie.

Comment 9

[rbs]

Attachment: transliteration for U+2212 - Official Unicode minus sign

Comment 10

[rbs]

It has been argued in n.p.m.mathml that <mo>-</mo> should render as \u2212 since
\u2212 is MINUS SIGN -- the official Unicode minus sign, and people usually use 
<mo>-</mo> to mean minus, not hyphen. By itself '-' is HYPHEN MINUS in Unicode. 
See http://groups.google.com/groups?hl=en&th=66488daf1ade7635&rnum=1

So I have hardcoded a substitution from <mo>-</mo> to \u2212 in the MathML 
code and this looks much nicer in math contexts indeed. However, as a fallback 
precaution, I would like to add a transliteration back to '-' to bullet proof 
against the typical problem of missing fonts.

shanjian, blizzard, care to r/sr? Thanks.
http://bugzilla.mozilla.org/attachment.cgi?id=70448&action=view

Comment 11

[Christopher Blizzard (:blizzard)]

Comment on attachment 70448 [details] [diff] [review]
transliteration for U+2212 - Official Unicode minus sign

rs=blizzard

Comment 12

[rbs]

Comment on attachment 70448 [details] [diff] [review]
transliteration for U+2212 - Official Unicode minus sign

Hoping to checkin as part of bug 126619 upon getting the clearance over there.

Comment 13

[Jerome Lacoste]

was this checked in with bug 126619 ?

Comment 14

[Jerome Lacoste]

ping! status update?

Comment 15

[Matthias Versen [:Matti]]

rbs: Is this now fixed ?

Comment 16

[rbs]

Nope. There are plenty of characters still missing. Needs somebody to comb the
list (comment 2).

Comment 17

[David Harvey]

FWIW, would be nice to map ⨯ to × if the glyph for ⨯ is unavailable.

See also bug 320062.

Comment 18

[rbs]

In transliterate.properties, we can only map a character to ASCII. For example, € becomes "EUR", − becomes "-" (hyphen-minus), © becomes "(c)".

We cannot map ⨯ to × because the transliteration processs doesn't support mapping to non-ASCII characters. If we were to do that in general, we would have to ensure the existence of the corresponding glyphs, and recurse if not. The added level of complexity would be excessive given that the transliteration is a last-gap fallback.

Comment 19

[Henri Sivonen (:hsivonen)]

Gfx has been rewritten not to need transliteration.