Konrad Rudolph

‘renv’ gets away with this for `install.packages` etc. — It doesn’t do this inside `.onLoad` but instead inside `renv::init` but that shouldn’t matter. This makes me more confident that CRAN...

I have to admit that I don’t entirely follow what you’re trying to do. I guess ultimately you want to know which packages/modules are loaded by ‘box’, right? There’s currently...

> Am I correct in understanding that the returned mod_spec object doesn't show what packages are loaded by a given mod? That’s correct, yes. The return value is an S3...

# Suggested API Module hooks are declared as special function names inside the module itself, analogously to R package [hooks for namespace events](https://stat.ethz.ch/R-manual/R-devel/library/base/html/ns-hooks.html). The special names shall start with `.`...

> Is there any reason to allow double definitions at all? Two reasons: 1. It reflects R’s builtin semantics: all symbols can be overwritten/shadowed anywhere. 2. Being the natural semantics,...

Ah. No, the stuff in `[…]` is the *attach specification*, not the import specification. It controls what names are attached to the calling environment (or, equivalently, exported directly).

Hmm. Apparently the answer is no: byte-compilation is not only not always faster, it can in some cases be _slower_. I’m still working out when exactly that’s the case but ...

Now that [package byte-compilation is default with R 3.5.0](https://cran.rstudio.com/doc/manuals/r-release/NEWS.html), this decision should be revisited.

How should this be compartmentalised? Should this be a core part of “modules”? I’m somewhat in favour of keeping it as a separate project. If so, would it be an...

It’s not solved already (none of this code is written yet) but I intended to implement this, yes.