Handle document created by parsing in XHTML. (mathjax/MathJax#2788) #758
Conversation
There was a problem hiding this comment.
I see that the solution works, although I find it rather broad.
So for future considerations, here are a couple of points:
- When using the
HTMLAdaptorI would assume it's behaviour to correspond to HTML. If there is no head or body something is wrong with the document and I (developer or client) would have to do something with that information. - Instead of conflating xhtml and html we could have an
XHTMLAdaptor. - We do have a
chooseAdaptorand abrowserAdaptorwhich could ensure that the right one is chosen depending on the document.
|
The issue occurs where a document fragment is parsed. In this case, it was the MathML that was being parsed as a string, like MathJax-src/ts/input/mathml.ts Lines 139 to 141 in 2dd53ce Would that be better? There are a couple of other places that might be necessary ( |
|
The If you wanted to use it for choosing between HTMLadaptor, the liteAdaptor (as it does now), and a proposed XHTMLadaptor, then any component that includes it would have to include all three adaptors, enlarging the component file for very little reason (there is no real reason to have the liteAdaptor in the browser, even if it is only 20K). |
|
I see and agree that we might not want to extend the |
MathJax uses HTML parsing by default for parsing MathML strings (so that when we support HTML-in-MathML, it should work). But in an XHTML document, the parsing structure is not the same as in HTML, so this PR alters the HTML adaptor to work with that structure as well.