should we separate marvin web from tools?

[havok2063]

While this goes against the whole principle of things, should we keep the official Marvin 2 web at Utah on master, or some other branch, while the pip-installed Tools on a release. This would allow us to potentially push web fixes and tweaks faster than the Tools releases. I'd prefer to have the web run on a fixed tag for some stability. Or should we just have people wait for releases to have web- and tool- fixes?

[albireox]

This is a good point but one that requires some discussion. In principle, I'm not opposed to making marvin-web its own product and allow for some independent development. That said, I'm not convinced that we cannot do that with the current system. A few points:

• We probably should remove marvin-web from the release distributions in PyPI (see #90). That would make marvin substantially lighter.
• We need to improve our deployment system (see #88) so that cutting a new version is almost pain-free.
• For releases that only affect the web maybe we don't need to release a PyPI version.

I think this would accomplish the same as having an independent repo for the web (especially if we are still going to have release versions for it, which we need). That said, we should discuss this more in depth.

[havok2063]

Let's brainstorm the idea of splitting marvin into 3 repos for web, tools, and api at the next Marvin Week. This has several advantages for the long term.