stan 2.5?

[Andrew Gelman]

Hi, just wondering when stan 2.5 will be released? With Cran and the diff equ models, this will be great.
A

[Daniel Lee]

I'm working on adding more tests for the ode stuff so we're reasonably confident the code will hold up. If anyone wants to help, let me know and we'll coordinate. Might take a bit of time to do this properly. (1 week?)

Bob's working on the documentation for Stan.
Everything else can wait.

I don't know RStan and cran status.

CmdStan will need a little bit of work, but should be straightforward.

PyStan seems to be up and running all the time. (Thanks, Allen!)

On Oct 2, 2014 11:46 AM, "Andrew Gelman" <gel...@stat.columbia.edu> wrote:

Hi, just wondering when stan 2.5 will be released?  With Cran and the diff equ models, this will be great.
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.

[Jiqiang Guo]

Ben knows better on the CRAN issue.  My thought is that we get 2.5 out first and then try putting it on CRAN.  There shouldn't be much work needed. 

Jiqiang 

[Allen Riddell]

All's well in the Python world. That said, I'm rather behind on C++ API changes.
If there's interest in releasing 2.5 simultaneously I'd need some advance
warning.

Best wishes,

-a

[Ben Goodrich]

On Friday, October 3, 2014 9:30:18 AM UTC-4, Jiqiang Guo wrote:

Ben knows better on the CRAN issue.  My thought is that we get 2.5 out first and then try putting it on CRAN.  There shouldn't be much work needed. 

I think the only thing we have to do is to remove the set_cppo() function and then just make sure the DESCRIPTION file is complete. So, you could tag 2.5 whenever you feel like the ODE stuff is tested.

Ben

[Ben Goodrich]

Is there some place where we can put -O3 so that it will get picked up when compiling models if the user has no ~/.R/Makevars file?

Also, too, making a rstanExamples R package now that there is a separate repo for examples. I had mostly done this already but forgot about it. The stan_demo() function lives in the new package now, so it needs to be taken out of rstan. I will make a pull request for all that.

Do you want to add a dependency on your cxxfunplus R package?

It wouldn't hurt to use two cores when building the vignette.

I see one new warning

In file included from ../inst/include/stansrc/stan/agrad/rev/var_stack.hpp:5:
../inst/include/stansrc/stan/memory/stack_alloc.hpp:17:9: warning: 'unlikely' macro redefined [-Wmacro-redefined]
#define unlikely(x) (__builtin_expect((x),0))
        ^
../inst/include/stansrc/stan/agrad/fwd/fvar.hpp:6:9: note: previous definition is here
#define unlikely(x) (x)
        ^
1 warning generated.

Ben

[Daniel Lee]

The new warning should be gone once the Stan submodule is updated to develop.

[Jiqiang Guo]

Please see below inline.  

Jiqiang

On Fri, Oct 3, 2014 at 12:41 PM, Ben Goodrich <goodri...@gmail.com> wrote:

On Friday, October 3, 2014 11:42:30 AM UTC-4, Ben Goodrich wrote:

On Friday, October 3, 2014 9:30:18 AM UTC-4, Jiqiang Guo wrote:

Ben knows better on the CRAN issue.  My thought is that we get 2.5 out first and then try putting it on CRAN.  There shouldn't be much work needed. 

I think the only thing we have to do is to remove the set_cppo() function and then just make sure the DESCRIPTION file is complete. So, you could tag 2.5 whenever you feel like the ODE stuff is tested.

Is there some place where we can put -O3 so that it will get picked up when compiling models if the user has no ~/.R/Makevars file?

I didn't find other ways to do it.  I think R_MAKEVARS_USER is only used for installing packages.  

Also, too, making a rstanExamples R package now that there is a separate repo for examples. I had mostly done this already but forgot about it. The stan_demo() function lives in the new package now, so it needs to be taken out of rstan. I will make a pull request for all that.

Great.  I guess we can remove the submodule defined in rstan repo as well. 

Do you want to add a dependency on your cxxfunplus R package?

No.  Back then, I put it on CRAN hoping other users would test it.  Now I want to remove it from CRAN, but I haven't bother to check how.

It wouldn't hurt to use two cores when building the vignette.

Are you referring to  http://goo.gl/u5aV0j?  It's two for non-windows. 

I see one new warning

In file included from ../inst/include/stansrc/stan/agrad/rev/var_stack.hpp:5:
../inst/include/stansrc/stan/memory/stack_alloc.hpp:17:9: warning: 'unlikely' macro redefined [-Wmacro-redefined]
#define unlikely(x) (__builtin_expect((x),0))
        ^
../inst/include/stansrc/stan/agrad/fwd/fvar.hpp:6:9: note: previous definition is here
#define unlikely(x) (x)
        ^
1 warning generated.

Ben