2.5 and beyond --- punting on higher-order autodiff again

[Bob Carpenter]

Since nobody said they're waiting for higher order autodiff,
we're going to punt again on getting it into 2.5 because it
involves an unknown amount of work on fixing forward-mode autodiff
test failures and nobody has any time to dedicate to it.

After the 2.5 release, our biggest priorities are going to be:

* getting higher-order autodiff done

* stiff ode solvers

Higher-order autodiff is blocking most of our research projects
and stiff ode solvers the rest of our research and consulting projects.

I'm also very keen to carve out some time for me to work on improving
the language. Mostly, I'd like to get in R's matrix indexing and
slicing, which I've sketched out an implementation for, but is going
to be a huge pain to code in both the parser, lvalue and rvalue semantics.
That and a compound declare/define statement.

- Bob

[Andrew Gelman]

This sounds great. Also let’s not forget to get rstan 2.5 on cran, this will help for various things!
Thx
A

> --
> You received this message because you are subscribed to the Google Groups "stan development mailing list" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to stan-dev+u...@googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.