Thanks Jeremy, that's very helpful, and it's great to hear from a developer!

I have been pursuing option (a), and I feel like I'm very close. The RUM tool runs, and the filesizes show up in the history, but the datasets show up as erroneous. When I click on the apparently-problematic dataset's bug icon, the error message shows the following two lines repeated over 100 times:

yes: standard output: Broken pip yes: write error

I know what this means, generally, but not in the context of Galaxy. Is this a telltale symptom, or is it too generic to say?

Under the additional output, it shows exactly the STDOUT the tool shows when it executes and terminates properly from the command line. So I know I'm close, I feel like I'm missing something small.

When I click on the "view dataset button," I see the data, and it's legit. When I click "Edit Attributes," I see a message at the bottom of the "Edit Attributes" pane that says " Required metadata values are missing. Some of these values may not be editable by the user. Selecting "Auto-detect" will attempt to fix these values." When I attempt to run the Auto-detect, this notification goes away.

It seems like the only issue right now is getting rid of that "broken pipe" error message. Once that's gone, perhaps the datasets won't be flagged as erroneous and I can use them in downstream processes.

If I can get this tool working perfectly, I'll definitely upload it to the Galaxy toolshed.

Any tips you could provide would be greatly appreciated!

Thanks,

Dan

You have two options:

(a) you can set up the tool to report only a subset of outputs from the tool;

or

(b) you can use a composite datatype to store the complete directory:

http://wiki.g2.bx.psu.edu/Admin/Datatypes/Composite%20Datatypes

Best, J.