You can't include the model_header.hpp file, as it's not
part of stan::math. Â Just try to include only the headers
that you need.
They declared everything as private rather than protected, not assuming
anyone would extend the damn thing.
Otherwise, we don't want to actually edit their class, you want
to fork it and fix it all in stan::math to be the way you want.
Their library lets you do that---you just have to keep their
copyright and mention at the top of the file that you edited it.
Hope that's not too much feedback from the get go!
Also, do not modify Stan's local copy of Eigen or Boost unless absolutely necessary because RStan doesn't use them. Just copy Eigen's unsupported matrix exponential file into stan/prim/mat and then tweak it.
--
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+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
File an issue with Eigen. They've already fixed a bunch of
stuff we've asked about, as has Boost. That's the beauty of
open source. Even better, submit a pull request for them (I have
no idea what version control system they use).
Are you on the stan-dev list? If not, you should be. I'm
cc-ing you just in case.
Sebastian
On Friday, August 19, 2016 at 2:13:12 PM UTC+2, Bob Carpenter wrote:
> I'm not sure if RcppEigen contains the unsupported modules.
> If not, we can just copy them into Stan so they'll work in
> all the interfaces.
>
> - Bob
>
> > On Aug 19, 2016, at 2:08 PM, Bob Carpenter
> >
> > Stan is unfortunately pegged to RcppEigen from R, so whenever
> > they update, we'll update. We're ready for 3.3.0, which will
> > hopefully be released soon.
> >
> > Looks like they use Mecurial (that version control system lost
> > the battle---nobody supports it any more), because that repo
> > says at the top:
> >
> > Git mirror of upstream Mercurial based Eigen repository.
> >
> > By "upstream", I think they mean changes to Eigen's Mercurial
> > repo get mirrored on GitHub, but probably not vice-versa. You
> > can see there aren't any pull requests or issues here.
> >
> > - Bob
> >
> >> On Aug 19, 2016, at 1:40 PM, Charles Margossian
> >>
> >> File an issue with Eigen. They've already fixed a bunch of
> >> stuff we've asked about, as has Boost. That's the beauty of
> >> open source. Even better, submit a pull request for them (I have
> >> no idea what version control system they use).
> >>
> >> They use GitHub: https://github.com/RLovelett/eigen
> >>
> >> Their latest stable release is 3.2.9, and Stan currently uses 3.2.8. The landscape in MatrixExponential.h looks quite different! Most importantly, not everything is private. I'll try to use the new version of code.
> >>
> >> Is it reasonable to expect Stan will use version 3.2.9 of Eigen?
> >>
> >>
> >>
> >> On Thu, Aug 18, 2016 at 4:39 PM, Bob Carpenter
> >> Are you on the stan-dev list? If not, you should be. I'm
> >> cc-ing you just in case.
> >>
> >>> On Aug 18, 2016, at 9:19 PM, Charles Margossian rote:
I'm not sure if RcppEigen contains the unsupported modules.
Yes, it is contained?
Stan is unfortunately pegged to RcppEigen from R, so whenever
they update, we'll update. We're ready for 3.3.0, which will
hopefully be released soon.
--
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+unsubscribe@googlegroups.com.
Stan is unfortunately pegged to RcppEigen from R, so whenever
they update, we'll update. We're ready for 3.3.0, which will
hopefully be released soon.Version 3.3.0 of what?Â
Bob and Ben have been working through the math library to get it to compile.
We're ready for 3.3.0, which will
hopefully be released soon.
It's all done, right? And isn't it Daniel that figured
out all the traits to make it work? Not that we're
keeping score.
- Bob
> On Aug 19, 2016, at 11:01 PM, Ben Goodrich <goodri...@gmail.com> wrote:
>
> On Friday, August 19, 2016 at 4:42:55 PM UTC-4, Daniel Lee wrote:
> Bob and Ben have been working through the math library to get it to compile.
>
> Bob has been working. I have been harping.
>
>
> --
> 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+unsubscribe@googlegroups.com.
--
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+unsubscribe@googlegroups.com.
> On Aug 24, 2016, at 10:16 PM, Charles Margossian <charles.margossian93@gmail.com> wrote:
>
> And to clarify, I mean the updates that make stan/math compatible with eigen 3.3.0.
Not sure --- Daniel?
- Bob