metadata after including .qmd files in other qmd files

[kris-brown]

Hi, when I include a file Y.qmd (with title "Y") via {{< include Y.qmd >}} into file X.qmd with title "X" (where title is given in the metadata at the top of the file) the title of X.qmd is rendered as "Y", as its metadata seems to be drawn from the final "--- [metadata yaml] ---" block seen in the file, after pasting in the contents of Y.qmd where the include statement was. A workaround is to just put the metadata of X.qmd at the bottom of that file, but is this expected behavior? It feels natural for the metadata for a .qmd file to come from the first "--- [metadata yaml]---" block that is seen.

[cscheid]

Yes, this is expected behavior. As described in the documentation, include works literally: it behaves as if the entire contents of the file were pasted at the site of the include shortcode. This means that if you include a file with metadata, you get multiple blocks, and pandoc interprets multiple blocks in this way.

The recommended workaround is to not include a .qmd file with metadata, but rather to include a .md file without metadata.

[kris-brown]

Thanks! It was natural to factor Y.qmd into Y.md and Y.qmd and then include the Y.md into both Y.qmd and X.qmd.

However, when quarto renders the Y.qmd, it seems like it automatically generates a Y.md for some intermediate computation and then automatically cleans up (i.e. deletes) Y.md. So one is precluded from ever having the same file name with both .md and .qmd extensions - this is a little unfortunate just because that convention would allow me to make it transparent "this is really the same file, but just wrapped with some metadata". My workaround is to just name the markdown file Y_.md but if quarto has any customizability to change how that intermediate file is named, that would be great to know!

[cscheid]

Oof. I'm sorry you got bit by that assumption in quarto.

Unfortunately, that convention is tied pretty deeply with quarto internals and it will be hard for us to change in the near term. I would recommend you use files starting with _ for includes, since those are automatically excluded from most of our wildcard calculations.