Merge this into gocraft/work

[austintaylor]

What do you think about merging this?

• gocraft/work has issues (e.g. Worker consumes a job more then expected gocraft/work#148)
• gocraft/work has 1.7k stars
• UV has been running your "prototype" in production for years

[taylorchu]

@austintaylor sgtm. but I don't have a good migration plan. do you have time to work on this?

[fgm]

If that is the only issue, it may not be such a problem: not having compatibility is part of creating a major version. As long as there is a doc detailing all the API changes (not just the new API), I would say it is fine, even though it is not optimal. Better done than perfect.

[ericraio]

just put the old stuff on v1 branch and merge this into master, we'll get a ton more support for this

[taylorchu]

Another major problem is that I don't have the time for answering questions on github. In old gocraft/work, there will be a lot of existing users asking about how to migrate. This is a primary reason why I didn't do that yet.

This library is mature enough: we run 10-50 millions jobs with this a day internally. It covers various workloads with custom middleware: some are 1hour+ long-running jobs, and some require extremely small latency like 100ms. It is very battle-tested, but it lacks good docs and examples. I think you can use this library now even it is not merged to gocraft/work yet.