714483 - MathML <lambda> not supported

Status: RESOLVED WONTFIX

(Core :: MathML, defect)

Description

[Gordon P. Hemsley [:GPHemsley]]

The MathML Test Suite has three tests for <lambda>:

http://www.w3.org/Math/testsuite/mml2-testsuite/Content/BasicContentElements/lambda/rec-lambda1.xml
http://www.w3.org/Math/testsuite/mml2-testsuite/Content/BasicContentElements/lambda/rec-lambda2.xml
http://www.w3.org/Math/testsuite/mml2-testsuite/Content/BasicContentElements/lambda/rec-lambda3.xml

All three are rendered incorrectly, compared to the "sample randering" and what the MathML spec says:

http://www.w3.org/TR/MathML/chapter4.html#contm.lambda

The spec offers up two possible renderings for "lambda x applied to sin(x+1)":

λx . ( sin( x + 1 ) )

or

x ↦ sin( x + 1 )

However, current nightlies render it as:

x sin(x + 1)

Bug 276028 comment 5 (from 2005) argues that content markup should not be rendered without corresponding presentation markup (within the same document?), but my skimming of the MathML spec and other MathML renderings suggests that that is no longer the general opinion (though I may be wrong).

Comment 1

[Boris Zbarsky [:bzbarsky]]

Gordon, we don't support content mathml in general, iirc.

Comment 2

[Gordon P. Hemsley [:GPHemsley]]

(In reply to Boris Zbarsky (:bz) from comment #1)
> Gordon, we don't support content mathml in general, iirc.

Well, it appears that <apply> is supported, if nothing else. Isn't that content MathML?

Comment 3

[Boris Zbarsky [:bzbarsky]]

I could just be recalling incorrectly!

Comment 4

[Gordon P. Hemsley [:GPHemsley]]

(In reply to Boris Zbarsky (:bz) from comment #3)
> I could just be recalling incorrectly!

Not entirely. Bug 276028 is, after all, for implementing content MathML. And a lot of the test suite data is consistent with the idea that content MathML is not generally implemented. However, I understand <apply> to also be content MathML, and that does appear to be implemented.

What I was mainly concerned about was the premise suggested in bug 276028 comment 5 that content MathML should not be rendered with a default presentational markup. Given that the aforementioned comment was made in 2005, as well as the fact that the spec offers sample presentational markup alongside just about every content markup sample, it seems to me that such an opinion is no longer the common one. Is that the case, or am I mistaken?

Comment 5

[Frédéric Wang (:fredw)]

Boris is right, we don't support content MathML at all (except <semantics/> which is considered presentation and content MathML). I don't know why you need content MathML, but you may be interested in David Carlisle's converter mentioned in bug 276028 comment 7, which gives quite good results.

Comment 6

[Frédéric Wang (:fredw)]

Mass WONTFIX for bugs related to legacy features that are not part of MathML Core (and with no plan to add them) or corresponding meta bugs. Please see bug 1495813 for the new meta bug.