"cabal fetch" should take output arguments

[gregorycollins]

It would be nice if "cabal fetch" could output the fetched tarball to a location specified on the command line, i.e.:

cabal fetch bytestring -o foo.tgz

This probably depends on #423.

[duncan]

Should this also save a copy in the normal local package cache or really just make the one copy at the target location?

[gregorycollins]

Just the one copy, methinks.

[duncan]

If it's already cached of course it's just a cp. In which case it may be simpler just to download it to the cache and then cp. Fewer different code paths.

[gregorycollins]

Good point -- I don't see any reason why that wouldn't be acceptable.

[duncan]

Also, if the output path is given as a directory then the file should be put into that directory with the obvious name, eg cabal fetch foo -o ./ would create foo-1.0.tar.gz in the current dir, or whatever version of the package it was.

[gregorycollins]

Setting priority on this to "high" because I may need it for platform.

[duncan]

And I should note that it's actually really very easy. Adding an extra command line arg is easy as is adding the plumbing to get the info to the right place. The only tricky thing is that for fetch -o we do not care about dependencies so we should use the logic for unpack rather than the normal fetch.

Bonus points for refactoring to unify the two methods and using a parameter to specify if we want to download deps into the cache too.