Open glwagner opened 4 years ago
Yes, I agree with @glwagner that it would be good to have an extra folder for our final paper results. It will be much clearer.
Yeah, a separate repo is even better
UPD: that said, I'd leave it as it is for now, and leave the issue open, and then maybe move it to a different repo before we open up, like suggested
I think it'd be forward-looking to move the code that is meant to produce data and figures for the paper to move to its own directory.
The purpose of
examples/
is to demonstrate how the code works. Its effectiveness is dependent on the smallness and simplicity of its contents. It is currently a bit bloated.Consider that the code in
paper/
will be frozen to a certain commit at the time of submission, even while the code continues to evolve. We will want to keepexamples/
working for the current version of the code, but we will not want to change the code inpaper/
, as it is meant to reproduce results reported in the paper.It could even make sense at some point to move
paper/
to a separate repository (this will make reproducing the results a bit easier, since we can record the relevant commit hash ofepiforecast
). It doesn't really make sense to mix a code (which is a living, evolving thing) with the scripts that are needed to reproduce paper results (which are fixed in time and depend on a specific version of the code / git commit). We do not have to decide on this point now, however; this could be the last thing that's done before publication and opening up of the repo.