Added ability to use functions at git's :commit option

[zilti]

Concerning issue #2482

Now, two questions:

1. Is this a good approach?
2. Should the function take an argument that denotes the VCS used so this change can be added to others like Mercurial, or are there better ways to figure out the VCS?

[zilti]

This is still work in progress. Actually, el-get-start-process-list is giving me some headaches. What would the preferred way be to use a return value from one process in the list in the next?

[npostavs]

2. Should the function take an argument that denotes the VCS used so this change can be added to others like Mercurial, or are there better ways to figure out the VCS?

Yeah, I think passing at least the type would make sense. Perhaps even the whole recipe, to be "future proof"? (might be overkill though)

Actually, el-get-start-process-list is giving me some headaches. What would the preferred way be to use a return value from one process in the list in the next?

I don't think you can use el-get-start-process-list that way.

[zilti]

I think passing the name and the method would suffice, since if the recipe is needed, it can be fetched using el-get-package-def.

I could modify el-get-start-process-list so that if an entry is a lambda instead of a list, that lambda gets passed the return value of the last process?

[npostavs]

I could modify el-get-start-process-list so that if an entry is a lambda instead of a list, that lambda gets passed the return value of the last process?

Hmm, a lambda is also a list, so this could get messy. How about starting with just a plain call-process and see about making asynch if needed later?