[x] Move non-core deps out of deps.sh + move stdlib into scannertools
[x] Make perf params required, add heuristic generation interface #243
[x] Make sure all the error messages are sane
[x] Improve error message when trying to write to a stream that already exists.
scannerpy.common.ScannerException: Per-stream arguments to op Sample must be a list.
scannerpy.common.ScannerException: Running this job would overwrite output 0 of op FrameColumn. You can change this behavior using sc.run(cache_mode=CacheMode.Ignore) to ignore outputs that already exist, or CacheMode.Overwrite to overwrite them.
AFTER MONDAY
[ ] Tensorflow Multi-GPU issue (Daniel's issue)
[ ] Guide[Extensions]
[ ] Guide[Kubernetes]
[ ] OS X build #233
[ ] Tensorflow kernel performance
[ ] Update kernels to annotate types
[x] Make it possible to get job id/profiler info without relying on sc.table
[x] Actually delete data in the database.
[ ] Update deps.sh to reinstall if updated dependencies
[ ] Move table methods to NamedStorage
[ ] Summarize method on Storage
[ ] Move all readers/writers into types
[ ] Python IPC #260
[x] List of standard library ops
[ ] Figure out why python scripts take a long time to exit
TODO:
BY MONDAY
Samplemust be a list.0of opFrameColumn. You can change this behavior using sc.run(cache_mode=CacheMode.Ignore) to ignore outputs that already exist, or CacheMode.Overwrite to overwrite them.AFTER MONDAY
sc.tabledeps.shto reinstall if updated dependencies