A windows version?

[einian85]

I think the title is obvious. I know there are problems with Rtools C++ compiler, but so what? what would happen? will you wait until some unknown day that Rtools will upgrade?

[eddelbuettel]

but so what?

I am not sure you really understand the issue. g++-4.6.* simply cannot compile the C++11 code used here. Which is pretty much what I try to say in DESCRIPTION.

So yes, if and when we have a new Rtools based on g++-4.9 (or newer) we will get RcppTOML for Windows. And no, I do know when that may be. There are efforts underway to get a new Rtools for R 3.3 next spring. But that was also planned for R 3.2 this spring ...

[krlmlr]

It seems that at least winbuilder is now capable of building the package on Windows (r-devel and r-stable). Is a CRAN re-release the easiest way to add Windows binaries to CRAN?

[eddelbuettel]

Details? How does one turn on the newer compiler needed for this?

[krlmlr]

I changed nothing but maintainer name and e-mail address in DESCRIPTION.

[eddelbuettel]

Can you point me to anything by Uwe that talks about when compiler is now used?

AFAIK Rtools3.3.exe is not yet finished; and the last comment by Duncan that I saw reiterated. So I'd urge patience.

I have a few other packages that would benefit from this...

[krlmlr]

...and I stand corrected: winbuilder doesn't throw up, but doesn't build binaries either. Drat :-)

Other people build C++11 packages using the new Rtools 3.3 toolchain on AppVeyor (CC @wush978). One might use that to publish binaries to drat, but this looks like it takes some effort.

[eddelbuettel]

And then you may still get into hot water for mixing different g++ versions -- though 4.6 to 4.9 should be safe.

[eddelbuettel]

Old thread ... but we just merged code that makes it possible to build on Windows.

Super excited, and nice work with @dpastoor to get there.