Radomir Stevanovic

Maybe it's enough to just link to it from the intro page, when you introduce the kerberos sampler?

Agree. Feel free to open a PR.

That would be nice. If all supported `compress` flag, we could use it by default from the client.

Benefits also heavily depend on network speed. On (very) fast networks, using anything except the simplest/fastest stream compression is probably not worth it. But we upload in parallel, so that...

Another alternative worth considering: [ultrajson](https://github.com/ultrajson/ultrajson).

When submitting 10 jobs to a 2000Q, runtime is ~120 s and peak memory usage ~800 MiB. When submitting 100 jobs to a 2000Q, runtime is ~13.5 min and peak...

@arcondello, I agree. Sampling params are just passed down to the QPU solver, and we can handle conversion there.

This basically translates to making `dwave.cloud.Client` picklable (in `dwave-cloud-client` package). Currently it isn't by default, because thread pools and mutexes would have to be recreated manually. Can you show here...

Blocked by https://github.com/dwavesystems/dwave-cloud-client/issues/217.

Not sure what exactly you have in mind. Something like this: ``` Client.cancel_all() ``` or ``` Client.cancel_problem() ``` or ``` Future.from_problem_id().cancel() ``` The last one of those I plan to...