I got a patch that makes this happen, but dependent on the changes in #3762 that in addition to deriving date from filename also fixes how metadata defaults are calculated which is needed to fix title handling. If that PR gets merged in I can easily make a PR for this issue too.

note I think its worth going a level further and try detect if content has a top level header as defined by markdown, asciidoc or orgmode and use that header if present, if not use filename.

note I think its worth going a level further and try detect if content has a top level header as defined by markdown, asciidoc or orgmode and use that header if present,

No, that is too hacky and not worth it.

No, that is too hacky and not worth it.

too hacky? as compared to try derive from filename ?

It works pretty well in awestruct and others and it has the big plus your blog posts actually render with proper title when shown "raw" in github.

btw. made the patch before I saw your comment so you can see the idea at #3859

note, the idea is similar to extract toc and summary that tries to parse raw and generated content to derive "metadata".

Note that the current TOC solution is also considered "hacky and slow" in my book, and will be fixed once Blackfriday 2 is out. We're not doing any regexp magic looking for titles.

Also, I'm not a big fan of pull requests that try to do more than the issue they describe. Do one thing well.

Also, I'm not a big fan of pull requests that try to do more than the issue they describe. Do one thing well.

okey - I've updated patch to just to title based on filename/slug.

I hope this can merged in together with the date fallback which this fix is dependent on (as without it metadata fallback calculation sequences are different whether there is a frontmatter or not )

How can I achieve this functionality?