Ingo Müller
Ingo Müller
The original version used IO control blocks from the stack, which will be overwritten after the function has returned. However, IO control blocks should live until the IO has completed.
This MR turns the project into a CMake project. The `compile-and-run-test` still works, but I removed the `Makefile` to avoid double maintenance. I have also made a small effort to...
It would be great to persist the history of the shell between invocations. The library you are using (?) hopefully has a configuration option for that.
I can't seem to import modules with starting Rumble with `--shell yes`. Running exactly the same query from a file using `--query-path` works. I have tried a couple of different...
I always get the following (running from the docker image of version 1.10.0): ```Warning! Some execution modes could not be set. The query may still work, but we would welcome...
With functions and modules that may be (transitively) called from several places in the query, it would be very useful to provide a stack trace instead of just the source...
The following query currently throws an error in my my timezone (Berlin/Zürich) ```xquery dateTime("1894-06-01T00:00:00.000") ``` After some digging, it turns out that this moment in time does not exist in...
The following should work even when Rumble is run with `--result-size 3`: ```xquery sum(parallelize(1 to 4)) ``` Currently, it fails with `Cannot materialize a sequence of 4 items because the...
It would great if `s3://` files worked also when run on Spark stand-alone. I am not sure what the right approach to do that is. Maybe the S3 connector can...
This would be great for benchmarking. The minimum set of metrics would be running time as reported in the shell. This isn't quite possible to get through the HTTP interface:...