naming of deploy / run / run-interactive goals

so I’m liking the fabric8:run-interactive behaviour (run and tail the log until Ctrl-C) be called fabric8:run as its then very much like existing goals developers are aware of (spring-boot:run, tomcat:run, jetty:run, wildfly:run etc). Seeing things start and the log is really really handy!

We already have fabric8:deploy / fabric8:undeploy to deploy/undeploy. I wonder if we just remove the current fabric8:run goal completely?

i.e. do we need a ‘run in background’ or ‘deploy’ goal that forces the Install goal to be run?

I wonder if these 2 options are enough:

mvn install fabric8:deploy
mvn fabric8:run

Or maybe we need some goal name for ‘install and deploy’; not totally sure what to call it though 😉

Maybe we rename the old mvn fabric8:deploy goal to be something more implementation specific; mvn fabric8:apply - then use fabric8:deploy to be invoke install and apply goals?

So folks could do either of these to deploy resources:

mvn install fabric8:apply
mvn fabric8:deploy

or to run interactively its

mvn fabric8:run

Then folks either use deploy/undeploy (which matches to other app server like goals) - then run is considered special (as its a deploy, tail log and undeploy on Ctrl-C combined into a single goal). Then things feel more app-server like?

[jstrachan]

@rhuss so I like the idea of folks being able to type mvn fabric8:deploy and things work (so triggering the mvn install goal before it).

Then we just need to decide if we need 2 flavours of this goal - one that does fork ‘install’ first and one that doesn’t. Personally I have found it handy once I’ve built an image to just invoke ‘mvn fabric8:resource fabric8:deploy goals’ - i.e. skip building java code, running unit tests + doing a docker build (which can be slow) - and just generate + reapply the manifests. e.g. if you do a build and the deploy fails due to some manifest issue; its nice to just rerun the resource generation + apply parts. Maybe for this advanced/internal goal we name it something more literal; how about mvn fabric8:apply-resources?

For the tomcat start versus run; it looks like tomcat has start/stop (which starts and stops a deployed app) and deploy/undeploy then run. Maybe we should do the same?

So start would scale the D/DC/RC/RS up to 1 and stop would scale it down again? (So its the less drastic version of delete?

On the cluster-start -> up - I kinda like start/stop for clusters as thats the commands that gofabric8, minikube, minishift use. Starting/stopping a cluster is more rare; but they feel like the right names so am still preferring cluster-start and cluster-stop really

[jstrachan]

gonna close this for now - lets reopen with new naming issues as we think of things we don’t like so much etc