Should we break the collector app into its own repo?

[groovecoder]

The collector app makes a good stand-alone system/service. If we put it into a dedicate repo, we could run it independently from analyzer code.

[peterbe]

I agree but I don't agree that it should be its own repo. At the moment, the app is completely independent from the analyzer (which technically should be renamed to redundantcss or something).

But I like that it's under one set of settings and that any of the plugins can just get to the collected html with from webalyzer.collector.models import Page

[groovecoder]

As I try to get this running on Heroku, it seems like a mis-match to have python/django collector, python analyzer (mincss), and node analyzer (crass) all mashed up in a single repo/system.

And the way the python code uses subprocess to run the node analyzers or tasks creates some tricky-to-debug issues. E.g., when the gator worker process tries to use a phantomjs process to generate a screenshot in the collected app.