Length limit for cross reference targets

[asmeurer]

There seems to be an undocumented length limit for cross reference targets. If I have something like

(aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa)=
# Header

where there are 36 as, It works fine, but if I instead have

(aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa)=
# Header

(37 as). Then it silently doesn't read (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa)= as a cross-reference target and just inserts it literally into the document.

Ideally I would prefer no length limit, since I like to make the cross reference target name match the header name, which can be longer than 36 characters, but if there is one, it ought to be documented somewhere, and ideally, give an error rather than just silently not parsing.

[chrisjsewell]

This is the link to the actual regex used: https://github.com/executablebooks/markdown-it-py/blob/c5f46e4bd95415d3b8483b3fa55c06318b25da71/markdown_it/extensions/myst_blocks/index.py#L8

I put the 36 character limit it for performance reasons, but can certainly extend it and document. It will still "fail silently" though, because it's not a failure in the parsers eyes; it just didn't match the regex.

Would 72 be ok, surely not longer 😬

[asmeurer]

Does the performance really suffer if you leave it unlimited?

[chrisjsewell]

Maybe not, but I'd prefer to keep a limit: 256 is my final offer 😉

[choldgraf]

@asmeurer can you share the use-case you've got where you need such long refs? Are you including entire file paths in the ref or something?

[asmeurer]

With recommonmark I was using direct refs like page.html#anchor. The anchor by default would be the text of the header. I decided to keep using anchor as the cross-reference name for MyST, since that is the simplest. For headers that have longer text, this results in longer anchors.

[chrisjsewell]

Ok so in markdown-it-py 0.4.8 (just released) I've increased the limit to 100, so hopefully that helps you out 😄

[chrisjsewell]

closed by #449