Done.

I don't think the issue has actually been merged.
Can you do that?

I think it's fine. There are some small things that should eventually be
addressed, like constructors that should be explicit and exception classes that
should behave more like normal exceptions, but that doesn't have to be part of
this issue.

If you see any problems with the code, Malte, we're happy to fix them in the
default branch.

We decided to merge this to get the haslums.

Code looks good to me. Malte, can we merge this?

Incorporated Jendriks comments.

A minimal build only for src/options and src/utils works.

I left two comments at Bitbucket.

I Updated Patricks PR. Can someone have a look at it?

https://bitbucket.org/cgeissmann/downward/pull-requests/10/issue873/diff

I also left some comments on Bitbucket.

I left some comments on Bitbucket and added a TODO to the summary.

The current PR can be seen under 
https://bitbucket.org/PatFer/downward/pull-requests/12/issue873/diff

But the outstanding issue844 and issue831 will add code which needs to be changed.

The option parser is mostly already planner independent. The planner dependent
part is once the

-commandline.cc separates the command line input and either forwards it to the
option parser or does its own stuff. The current idea is to move the
commandline.{h,cc} one directory up and out of the option parser.

-the option parser uses some code from utils. This is in general fine, as most
utility stuff is planner independent. Currently we are only aware of the Exit
Codes which are planner dependent chosen. Ways to get around  that are:
Require one specific ExitCode (e.g. OptionParserError) to exist, throw an
exception and the planner has to handle it (currently favored)