lifegadget / aws-pusher

An automated deployment tool for static assets to AWS
0 stars 0 forks source link

We aim for cli so we don't need to download ember-cli #4

Open BryanCrotaz opened 9 years ago

BryanCrotaz commented 9 years ago

But a push requires a build, so that requires ember-cli. So back to my previous point, why build something outside the cli framework?

yankeeinlondon commented 9 years ago

Previous point? Sorry not sure which point you're referring to but we talked about this on the phone yesterday and I thought we were in agreement to keep Pusher outside of Ember-CLI for now.

As to your point on push requiring a build ... I agree they are fairly coupled but having a little separation for me is not a bad thing. Example:

yankeeinlondon commented 9 years ago

So in short:

Make sense?

yankeeinlondon commented 9 years ago

I'm not as familiar with git hooks as I should be but we'll likely need a few hooks to pick up all events we're looking for.

BryanCrotaz commented 9 years ago

I thought the reason for going to a "pure" cli was so that we didn't need to download ember-cli, which is huge.

I don't think ember build should call into pusher push, but pusher push should call into ember build, so they are tightly coupled. In that case, why not make this an addon? I can see the point of doing a pure cli if they are loosely coupled, but they're not.

On 23 April 2015 at 11:09, Ken Snyder notifications@github.com wrote:

I'm not as familiar with git hooks as I should be but we'll likely need a few hooks to pick up all events we're looking for.

— Reply to this email directly or view it on GitHub https://github.com/lifegadget/aws-pusher/issues/4#issuecomment-95523563.

Bryan Crotaz Managing Director Silver Curve

yankeeinlondon commented 9 years ago

The reason for pure CLI is that there is very little gained from the direct connection (unless I'm missing something) and by not being purely tied to Ember we're opening up to a potentially much larger audience (aka, Angular, and any other static build solution). I'm not directly concerned with that larger audience but I just don't see a big benefit of tying to Ember-CLI as a state engine when Git is the bigger driver of events. I still hold that this is not tightly coupled but WRT to the build-to-deploy cycle of the developer sandbox there's no getting away from the relationship (just like there's no getting away from the commit-to-deploy relationship for commits, tags, and branches).