feature: support overriding host name on au run, and e2e tests

[shahabganji]

This pull request helps you to override the default running host on webpack configuration from localhost to whatever you serve the au run command. This will help you to satisfy both #1122 and #1130

e.g.

au run --host 127.0.0.1 --port 8090 

au run --host 0.0.0.0 --port 8090 

and you can still have the --open command line argument as well to satisfy #1130

au run --host 127.0.0.1 --port 8090 --open

or simply au run --open to run the application on localhost:8080

---

There was a minor misbehavior with --open on my machine, it opens two tabs, that is also fixed in this PR.

[3cp]

Please update skeleton/webpack/aurelia_project/tasks/run.json too for the new "host" option. The file is used by au help.

[3cp]

Looks great. "open" can be removed from cli package.json too.

[shahabganji]

@3cp, the CLI bundler part is tested with Alameda, if further testing is required let me know, otherwise remove the WIP flag at the beginning.

If webpack has been selected as the bundler, we can override both port and host

au run --port 8081 --host 127.0.0.1

if CLI bundler has been selected, only port can be overridden, I didn't find how to override the running host, if you know the place, let me know.

au run --port 6060

For the dotnet core template we could override the port via the --port flag, however, for the host it is kinda useless to override it. If you think it should be overridden as well, let me know of a mechanism that I could read the values within .ext files.

[3cp]

browserSync supports host option too, you can update in the run task.

[shahabganji]

@3cp

karma-coverage > [email protected]: This module is no longer maintained, try this instead:
  npm i nyc
Visit https://istanbul.js.org/integrations for other alternatives.

should we change something for this warning?

[3cp]

Not in this PR pls :-)

[shahabganji]

@3cp,

Which one not in this PR, this commit, or updating istanbul to nyc?

[3cp]

Yes, I mean outdated istanbul. Replacing it looks like not a small change.

[3cp]

We have merged and tested the webpack refactoring PR.

The run task has been simplified to a wrapper of npm start.

You can still pass au run --host x.x.x.x --port 8888 to au run, the additional parameters are handled by webpack-dev-server directly.

You would need to rebase on top of current master and resolve the conflicts.

[shahabganji]

@3cp
How can I gracefully stop the webpack server? programmatically, like the shutdownAppServer we exposed in this PR.

[3cp]

Probably can do from outside.

I updated release-check to use tree-kill to kill au run (which wraps webpack-dev-server command). It might be useful to you too.

Probably you can use cross-spawn (for reliable win32 spawn) to spawn au run or webpack-dev-server directly, hold the proc instance, then use tree-kill to kill proc.pid.

[3cp]

Or you can skip au run and webpack-dev-server command, but use webpack-dev-server API. But I remember you pointed out a bug that prevent the dev server API for close to work.

[shahabganji]

But I remember you pointed out a bug that prevents the dev server API for close to work.

That was for browser-sync.

I am not sure about tree-kill I'll give it a try, but what was weird to me is that when I killed the process with either process.kill(pid) os ps-node 's ps.kill(pid), it looks like they killed the process, but the port is still in use, either unit tests passed or failed.

[3cp]

but the port is still in use, either unit tests passed or failed

This is what tree-kill solved in release-check. It seems normal kill will skip some child process.

[shahabganji]

And one more thing, how could we tell tree-kill to fail the processes with a proper exit-code when tests failed?

[3cp]

I don't get your question.
tree-kill is for the running dev server process. If dev server failed to start, you should have proc.on('error/exit',cb) to handle it. Only when dev server started properly (by detecting stdout message like release-check task), you start e2e tests, when e2e tests finishes, you tree-kill dev server, and use the result of e2e tests to exit, not the result of tree-kill.

[shahabganji]

@3cp
I think it's done. Check it out and let me know of any ideas or feedback.

[3cp]

Seems you got lots of duplicated commits/changes with those merges. Do you mind to start clean with another branch on top of latest master? New PR is if needed.

[shahabganji]

@3cp check out #1143, I close this one in respect of #1143 😄