Peter Hill

I think the example in #366 should be a sufficient test case. Unfortunately I don't have a massive amount of resource to spend on ford, so it might be some...

I think the cleanest approach is to base it on the existing [admonition](https://python-markdown.github.io/extensions/admonition/) extension. This is probably a breaking change though, as I think it would involve dropping support for...

Rendering to markdown might also make it easier to use e.g. Jekyll themes too (see also #280). I'm reopening this, it seems like a useful thing to do if someone...

I'm afraid there's currently no mechanism for dealing with this. When the files are preprocessed, the text of the `#include` files is literally included, so ford has no way of...

Maybe there's a dockerhub token we can add, similar to the PyPI one?

Yes, we should definitely try and save the complete settings as soon as possible. `OptionsNetCDF` does some of this already: https://bout-dev.readthedocs.io/en/latest/user_docs/bout_options.html#reading-and-writing-to-netcdf That uses groups, which I think is the more...

I'd noticed the build times, but not the library increase. That's pretty big! Even 400MB seems like a lot. I suspect I messed up forwarding the parameter pack. Suspect it...

Also I include `fmt/format.h` everywhere, when `fmt/core.h` should be sufficient almost everywhere. I've pushed a branch to see if this helps on Travis. But neither changing the header nor forwarding...

Doesn't look like that made it any faster :disappointed: https://travis-ci.org/github/boutproject/BOUT-dev/builds/664940917

Interesting article on compile times: https://cor3ntin.github.io/posts/compiletime/ I thought I had made an attempt at using ccache on Travis, but I can't find it now. It might be a branch on...