2015-06-30 11:04 GMT-05:00 John Chilton <jmchilton@gmail.com>:
On Mon, Jun 29, 2015 at 2:48 PM, Alexander Vowinkel
<vowinkel.alexander@gmail.com> wrote:
> I have a tool with filtered outputs (<data [..]><filter>[...]</data>).
>
> A) As I understood the filters are pure python. But if I put it in
> "<![CDATA[", it doesn't work. I think this should work even better ;)

This sounds wrong - the standard library is handling the CDATA parsing
for us and so Galaxy should be looking at an identical representation
of the data. Are you sure you are using the CDATA tags correctly? I
modified the test tool demonstrating output filters with the following
patch (https://gist.github.com/jmchilton/1c8d13c01815a9b0fc8e) and ran
the tool tests using the command:

./run_tests.sh -framework -id output_filter

and everything seemed fine.

Yeah. Weird. Now I don't have problems anymore too.
Must have been something else.
 
>
> B) The workflow generator doesn't interpret the filters.
> When I put the tool in a workflow, all possible outputs are shown. This
> doesn't change when I change an option that would change the output choice.
>
> Can someone fix this? ;)

This would be a nice enhancement - but it would take a lot of effort
to implement. I would suspect we are going to be stuck with the
display of invalid outputs for a while yet. I've created a Trello card
for this issue:

https://trello.com/c/Hcm2MlQg

What a pitty :(