johnmchambers

Correct. The package was tested vs Julia 0.6. The plan is to update it. Any suggested changes would be appreciated. On Dec 6, 2018, at 5:41 PM, olugovoy wrote: Hello,...

Thanks for the catch on fieldnames(). Should be fixed as of commit today. The convert() warning doesn't seem to come from XRJulia. No use of this Julia type in the...

The immediate problem is that any object sent directly to Julia uses JSON representation, which will be bulky and inefficient for large arrays. It's unfortunate the Julia parser isn't more...

Of course. It still has to go through JSON. The point is that once you *can* send the large object, you should do it once and then refer to the...

The version of XRJulia now on Github (version0.7.9) now uses binary reads for numeric or integer vectors of size > 1e4 bytes, including when they are part of a structure....

Works OK for me with the CRAN version of the package, running under MacOSX 10.12: > x2 = juliaEval("[i for i in 1:1001]") > juliaGet(x2) [1] 1 2 3 4...

You're right though. It's a bug introduced in the modifications for 0.7.9. I'll try to track it down, but not for a few days. Thanks for the report.

Turned out to be an easy fix; for the new fast transfer of data, the Julia side needs to convert Int64 arrays to Int32 before writing. Also incremented the version...

If I understand your R script, it's not juliaGet() per se, but the fact that you are running MCMC (which is not fast in R anyway) with a 2-way interface...

Vaclav's summary is very likely the point. Rcpp has two speed advantages over XRJulia in terms of general interfaces (e.g section 12.3 of Extending R): it's at the subroutine level...