hey Ted On Mon, Nov 21, 2016 at 8:20 PM, Ted Dunning <ted.dunn...@gmail.com> wrote: > Wes, > > This is awesome. > > Does it, however, imply that to run the tests that a C programmer will need > a working Java environment and a Java programmer will need a C environment? > > Is there any way around that? Possibly by storing golden bits for the > in-memory images somewhere? >
Easiest thing would be to create a Dockerfile for experimentation -- this would be useful for benchmarking on different hardware environments as well. We'll want to run the integration tests either in Travis CI or Circle CI anyway (right now we have the Java and C++/Python unit tests running in separate build setups in Travis CI), so it hopefully wouldn't be a great deal of additional effort to put everything into a container recipe. Wes > > > > > On Mon, Nov 21, 2016 at 2:05 PM, Wes McKinney <wesmck...@gmail.com> wrote: > >> hi folks, >> >> After a long road, we're getting very close to having tests proving >> that the Java and C++ Arrow implementations are binary compatible -- >> this will be an exciting major milestone for the project. If you >> haven't been following along recent JIRAs, the way these tests work is >> as follows: >> >> 1) Testing dataset is specified in JSON format >> >> 2) Producer library (e.g. Java) reads JSON into Arrow in-memory, then >> writes out to an Arrow file IPC binary format >> >> 3) Consumer library (e.g. C++) attempts to read both the JSON and the >> binary file yielded by the producer library. The consumer compares the >> in-memory schemas and columnar data structures and indicates whether >> they are binary-identical >> >> I found a couple initial incompatibilities in the file format >> implementations, cited here: >> https://github.com/apache/arrow/pull/211#issuecomment-262080545. >> >> Thanks >> Wes >>
