Add post-checkout/clone entry point for custom code

[reinout]

See https://bitbucket.org/reinout/checkoutmanager/issues/23/allow-an-user-defined-init-command-to-run

[reinout]

Original question by @do3cc:

I'd love to have the option to define a script that shall be called after a fresh checkout. I use virtualenvwrapper, and when checkoutmanager checks out a new repo, my first action always is to rename the dir, create a project with the same name and copy the renamed contents into the new project dir.

I've now created a script for myself and it contains a lot of special logic already that is not virtualenvwrapper specific: https://gist.github.com/do3cc/520c9b0b03466f2bf469

So I think the option is the best way forward, but maybe you have an even better idea. Else I create a PR

[reinout]

(Copy of my original answer from the bitbucket issue:)

Having checkoutmanager call something after a checkout? Nice idea, sounds useful. It also sounds like a "post-checkout" command, which suggests that someone might like a "post-update" command

What do you mean by "option"? Something like "checkoutmanager co --co-script=/usr/local/bin/something.sh"? Or do you want a special "[checkoutmanager]" entry in the config file with one or more options?

Taking the last idea (I'm brainstorming...), checkoutmanager could look for a "post-checkout/post-update" entry in every section, which means you could swich on something like this just for specific kinds of checkouts. With a fallback to a default in "[checkoutmanager]".

Alternatively, we could use setuptools' entry points. I have that in zest.releaser (http://zestreleaser.readthedocs.org/en/latest/entrypoints.html) and it works like a charm. People use that to customize zest.releaser's behaviour. It is a great way to get rid of the more outrageous pull request: "just write it as a custom entry point". :-)

The advantage: you don't need to pass any option. If you want something custom, just install some package and checkoutmanager will automatically pick it up.

Also an advantage: because you're calling a python function, you can easily pass a bunch of arguments. Directory name, checkout url, options in the section (possibly useful to configure the entry point...).

This brings me to another point: do we want to call a script (python/shell/binary/whatever) or do we want to call a python function? Personally I'd now go for the python function in combination with an entry point. Nothing stops you from just calling a shell script in the python function, of course.

What are your thoughts?