Andrei Zhabinski
Andrei Zhabinski
According Apache Arrow description, it provides "zero-copy streaming messaging", which may be helpful to eliminate the cost of transferring data between JVM and Julia. Julia already has [bindings](https://github.com/ExpandingMan/Arrow.jl) for Arrow,...
Originally, Spark.jl was developed for Spark 1.x with its RDD interface, e.g. most useful functions were `map`, `filter`, etc. In Spark 2.0, the default interface changed to `Dataset` (`DataFrame` in...
During some refactoring we have silently broken some `PairRDD` methods. Broken methods can be seen in [tests](https://github.com/dfdx/Spark.jl/blob/5c3579b26ea063217899074abc4dee3979427451/test/julian_versions.jl).
See #62 for details.
Register
An issue for the registration bot
Current approach to dynamic graphs is to trace function on each execution and select matching tape (with diff operations already added) from cache. This approach has several disadvantages: 1. Tracing...
Support simple forms of array comprehension, throw a meaningful error for more complex forms.
IRTools provides line info in `ir.lines`, so it should be possible to pass this info to `Call` nodes in the `Tape`
`jcall` fails with `NoSuchMethodError` when calling methods of a dynamically compiled class. I use [InMemoryJavaCompiler](https://github.com/trung/InMemoryJavaCompiler/blob/master/src/main/java/org/mdkt/compiler/InMemoryJavaCompiler.java) to compile a new class and create its instance. (`InMemoryJavaCompiler` is based on `javax.tools.*`, so...