Pass parsed metadata into pre_knit

[cderv]

They will be read before pre_knit and access can be useful to do some side effect based on metadata in file. (use case in rticles)

Current workaround is to re-parse the YAML from the input file using rmarkdown::yaml_front_matter()

This is easy enough though, so maybe it is ok to keep it that way.

I opened the PR to share the idea as we always told in our doc that pre_knit should support ... for future expansion, and it seems ok to add

What do you think ? Did I miss something that would prevent us doing this ?

[yihui]

Sounds good! My only concern is whether all people have really followed the advice in the documentation to add the ... argument to their pre_knit function. We'll know when we do the revdeps check.

[cderv]

My only concern is whether all people have really followed the advice in the documentation to add the ... argument to their pre_knit function. We'll know when we do the revdeps check.

yes I have the same concern. We'll see.

[atusy]

Maybe we can test if ... is available.

If not, we can fallback to the previous behavior with warning.

names(formals(function(x, ...) x))
#> [1] "x"   "..."

[atusy]

This warning will also be helpful when we add more arguments in the future.

[cderv]

@yihui did you ran a reverse dependency check on that yet ?

I thought we would try merge it before next release and see if all passes but I think you have done the release already.

I'll run one on this for next release if you haven't done it.

[yihui]

Let's merge this after the next release (which is on its way). Thanks!

[yihui]

Sorry I forgot this PR. I just started the revdeps check: https://github.com/yihui/crandalf/actions/runs/6198690012/job/16829628626 I'm hoping to make a new release in the next few days if this PR works for revdeps.

[yihui]

Revdep checks are finished. Only one package would be broken: https://github.com/line/conflr/blob/74c3c36f0a786138d4d90f49df45ab165901761c/R/document.R#L133 Good news is that this package has been retired. Anyway, I'll use @atusy's idea above to test for the existence of ... in case we'd break any other packages not covered by our revdep checks.