Fork me on GitHub
#onyx
<
2016-12-02
>
lucasbradstreet04:12:30

Hi all, 0.9.15 is out https://github.com/onyx-platform/onyx/blob/0.9.x/changes.md#0915. A few defaults that have caused issues in the past have been changed, and we have reduced the number of dependencies in onyx. One important addition is an aeron media driver health check that has been added to onyx-peer-http-query https://github.com/onyx-platform/onyx-peer-http-query/blob/master/CHANGES.MD#09150 cc There’s also a new after-apply-fn lifecycle that is after the onyx/fn is applied, but before the batch is written

yonatanel15:12:32

@lucasbradstreet Sent you a PR for a minor fix in the new docs

michaeldrogalis20:12:00

Rich’s talk from the Conj is quite provoking. We’ve had a lot of conversations over the years about Onyx being sort-of-mostly-I-tried-hard semantically-versioned. Since we’re nearing 1.0, maybe it’s time to abandon that.

Travis20:12:33

Crap, i forgot the conj has started!

Travis20:12:39

miss not being there

michaeldrogalis21:12:00

Timestamped versions + specs indicating incompatibilities would certainly make our release tooling easier. There is a lot of infrastructure that makes sure when we release Onyx w.x.y, the plugins release w.x.y.z, while that seems appealing at first, it’s clearly not been free from problems - on both ends, for us as the producers of the artifacts occasionally making incompatibilities in the wrong places, and for consumers not remembering to lock-step versions

michaeldrogalis21:12:53

It’s especially interesting in our case because another thing that we do behind the scenes are automatically run tests for all plugins everytime there is a push to master. But the reason we’re able to do that isn’t because of semver, it’s because we specify dependencies as Git SHAs for everything except proper releases - meaning there’s no incongruity for our test process if we stepped away from semver.