Github user superbobry commented on the issue:
https://github.com/apache/spark/pull/21157
Nope, the job I was referring to is not open source; but I guess the
speedup is easy to justify: much less payload and faster deserialization:
```
>>> from collections import namedtuple
>>> Stats = namedtuple("Stats", ["sample_mean", "sample_variance"])
>>> import pickle
>>> len(pickle.dumps(Stats(42, 42)))
31
>>> len(pickle.dumps(("Stats", Stats._fields, (42, 42))))
68
```
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
