Makefile upgrades. Can anyone review them?

The makefiles across Math, Stan, and CmdStan have all been updated. They’re code complete, PRs are open, and are awaiting approval. They’ve been vetted by a number of different developers (each testing a subset of platforms, hopefully covering it all).

Here are the PRs / issues:

The status will show as not passing all tests. This is expected: Travis won’t pass until the PRs are merged. Jenkins is showing as having passed (we update the appropriate submodules to ensure it).

Hopefully the issue itself isn’t something up for debate (that the makefiles could use an overhaul). So what remains is taking a look at the makefiles and determining whether it’s better than what is currently on develop. Ideally, it’ll be easy enough to know how to extend it directly as someone that doesn’t really use make that much. But we’re looking for an incremental improvement, which I think this provides. I think it’d be great if we could move off of make completely, but that’s a different effort.

ccing the people that have been involved so far: @wds15, @rok_cesnovar, @mitzimorris, @bbbales2, @yizhang, @sakrejda.
ccing additional people that could take a look: @seantalts, @Bob_Carpenter.

@mitzimorris is doing it.

Thanks, @mitzimorris and @wds15!

Thanks @syclik for taking on this task. Having gone through the pain of adding MPI to it, this new system should avoid a lot of future hurdles for others.

(and on top we gain much improved dependency management)

1 Like