Clojurians Log v2
clerk
Since upgrading to latest clerk, some of the pages in my project no longer load, with an exception:
Clerk evaluated '/Users/howard.lewisship/workspaces/nubank/nubank-work-activity/notebooks/nu-dev.md' in 76.420708ms.
Execution error (ExceptionInfo) at nextjournal.clerk.viewer/viewer-for (viewer.cljc:1458).
cannot find viewer named :nextjournal.markdown/html-inline.md file, just contains some basic free-form notes in markdown.
This is in version 0.18.1150.
can you try if the error persists after calling (nextjournal.clerk/clear-cache!)
do you have an explicit dep on nextjournal/markdown in your project?
No, don't have such a dep. clear-cache! doesn't help, nor does restarting my REPL.
Nothing in the stack trace is helping me track down where in my document Clerk is having a problem. I'd share the doc but it has a lot of Nubank internal stuff I can't share.
ok, I can repro it now
Awesome! What triggers it?
inline html 🙂
hello
<strong>world</strong>Ok, I don't have that in the document though? Just ordinary markdown.
strange, need to investigate more then
Found it:
nu service channel <project>it looks like we added inline html support to nextjournal/markdown and didn’t add the corresponding viewers to clerk
will fix it and cut a bugfix release
thanks for the report!
Cool. I see this as one part user error, one part error reporting failure (tell me the line where things failed, please!)
true, that would be helpful context for sure
created https://github.com/nextjournal/clerk/issues/762 to track this
not sure how easy we can surface the error location in the markdown
it’s somewhat unexpected that parsing a normal markdown doc results in a viewer error from clerk
Should be fixed on the main branch now. We're just passing through <project> as plain text
@hlship you weren't actually intending to use inline HTML right?
No, that was not my intention. Indented with four spaces, it was treated as a literal plain text block containing some brackets, as I intended. But with two spaces, it was still markdown that tried to interpret <project> as (invalid) HTML.