com.kotcrab.remark:remark will be archived

[rgrunber]

As per kotcrab/remark-java#4 , https://github.com/kotcrab/remark-java will be archived.

It's also used by the XML language server to convert HTML to Markdown. Are there any actively maintained alternatives or will we end up just forking it in some other repo ?

CC @testforstephen , @fbricon

Update : Maybe https://github.com/IBM/MDfromHTML ?

Update 2 : remark depends on jsoup, which is vulnerable to GHSA-gp7f-rwcx-9369 . However, in our case, Javadoc don't use javascript: do they ?

[fbricon]

We should explore alternatives, I don't really want to maintain a whole other library. https://github.com/vsch/flexmark-java might be interesting. Others might fit too.

[fbricon]

MDFromHTML likely doesn't have @NikolasKomonen's fixes. You'd have to check and port them if necessary.

[datho7561]

I tried using flexmark-java in lemminx:

• It's not a drop-in replacement; there were some differences in how the library handled links, and sometimes it would add whitespace at the end of lines. This meant the unit tests failed. However, the difference isn't that noticeable from within VS Code, so I think we could just modify the unit tests to the new behaviour.
• The lemminx uber went from 8.2MB to 9.5MB; it looks like the library with its dependencies is around 1.5 MB
• It looks like we were using reflection to modify the settings jsoup used to clean the HTML; it wasn't obvious how we would do this using flexmark-java

[fbricon]

My experiments using flexmark in lemminx are rather positive so far, except it doesn't run in binary mode yet.

Anyways, this doesn't concern jdt.ls, so I will port my code here.

It's annoying it now adds ~4MB of jars, but there's no need to use some reflection hack to configure JSoup anymore.