Hi all, 
I was recently wondering how it might be possible for 
an output from tool A to be used as a parameter when 
calling tool B (without modifying tool B to take a data 
configuration file or anything like that). 
To clarify, I'm thinking about simple parameters like 
a single integer/float/string like "median insert size" 
or "average coverage" which map to a single input 
parameter in a downstream tool. 
Workflow parameters basic $variable support is 
similar to what I'm thinking about here, see: 
https://wiki.galaxyproject.org/DevNewsBriefs/2011_01_31
Perhaps one day in addition to tool outputs being files, 
they can include setting $variables (local to the history), 
which could then be used within downstream tools? 
I can see this has downsides with tracking the providence 
of each variable though... thus: 
Another idea would be extra filetypes for each parameter 
type (Unix philosophy - make everything into a file), e.g. 
integer_parameter, float_parameter, ... which "Tool A" 
could define as an output, and write to. Galaxy would 
then need to offer these history entries to the user when 
a later tool (e.g. "Tool B") asked for a parameter of that 
type (e.g. an integer, or a float). 
(This idea could be extended further, with data files 
using JSON or something to hold a structured set of 
parameter values which Galaxy could then offer to 
fill later tool's inputs.) 
Crazy, or is there some mileage in this idea? 
Peter