Dear all,
We use the following query to see if a dataset is shared between more than one history: select count(*) from history_dataset_association hda where hda.dataset_id=NNNNN
And the following to see if it's shared between more than one user: == select count ( distinct history.user_id ) from history, dataset, history_dataset_association hda where dataset.id = hda.dataset_id AND history.id = hda.history_id AND dataset.id = NNNN ==
Thanks a lot for these helpful scripts. I've just tried some of them out and they work wonderfully.
Are there other types of sharing (between users) ?
You can gererally publish a history, so anyone can access it.
I'm not sure about the best UI way of displaying those purged datasets, but something like showing the datasets just like the current green rectangles (or the way deleted datasets are displayed, with a warning), showing the user what was the analysis, the tools, the parameters, and if/when the user clicks on the "eye" icon or the "download" icon, he will get a message saying "this dataset has been purged".
--- or an option to rerun all the tools to get this dataset again.
But he will still be able to "re-run" or view the tool's parameters.
The rational behind this: I'm in the same situation as Sebastian - I have users running big jobs, on many FASTQ files, on crazy datasets (example: Genomics Interval's "Join" on two 300MB BED files, producing a 70GB file, then filtering the results with GREP, or 8 x PE100 FASTQ files that go throw the same workflow). Many times they don't need the intermediate files, but never bother to delete them (this was before the workflows had an option to delete intermediate files, and even now not many people are using this feature - and I can't force them).
This could be the default... especially if the GUI provies a way to "restore" or re-generate the intermediates
Having a way to display the meta-data of a purged dataset (especially since each dataset has a "peek" and "info" data) would be very helpful (helpful - but not top priority - I don't want to create the wrong impression).
True, I'm sure there are more pressing things. However, you could put it on the nice-to-have list or do a minimal implementation like the ones that show up like deleted datasets. An expiration date may be very hard to implement. Thanks again! -- Sebastian