Bug fixed: when uploading a gzipped FASTQ file through upload1 tool on a
machine with SQLite 3.3.6, it was detected as 'sqlite' datatype and not
decompressed.
In fact, the sqlite3 command 'pragma schema_version' on any file returns
without error on this SQLite version.
Bug fixed: when uploading a gzipped FASTQ file through upload1 tool on a
machine with SQLite 3.3.6, it was detected as 'sqlite' datatype and not
decompressed.
In fact, the sqlite3 command 'pragma schema_version' on any file returns
without error on this SQLite version.
Tools will pair off (or triplize, quadraplate, quintukify, etc...) based only on the index of the elements in the collection (so any N element list can be used in the same tool execution as any other N element list). Originally, I had decided to require the element identifiers to be the same in both collections (allows mixing a matching fewer collections in tools - but probably less likely to mismatch in a way the user did not intend). At any rate - I changed that decision and updated tool execution appropriately but not every aspect workflow collection 'pairing' - this fixes that and closes out this card https://trello.com/c/m5PiF1lR.
In very partical terms tools would ideally be consuming collection pairs - but for those that are still consuming two separate files (i.e. all of them) one can use two lists of datasets with pairs at matching indices and feed them into these tools - for workflows this doesn't work without this patch unless the identifiers and the indices match on these collections.
Thanks for Michael Crusoe and takadonet for the bug reports.
Web form (at least for older style upload tool) was passing in '' and the UUID custom datatype treats this as an invalid UUID (probably correct) instead of a None. This changeset fixes that by ensuring these values coming in through the controllers are transformed into None-s.
Move all operations prior to workflow invocation (and a few other places state and modules are used) relating to populating the non-persistent attributes of `WorkflowStep`s (e.g. module, state, upgrade_messages, and input_connections_by_name) into a single function on a new class.
This should fix a couple of small errors as well - input_connections_by_name was being mis-calculated in a few places (though potentially never used down those roads) and some error handling code in the workflow editor template was expecting step.module to be populated with something (even if None) and it was not always.
See full conversation on https://bitbucket.org/galaxy/galaxy-central/commits/07508592cb65845474b9f3e0f781260c.
Would like to better and allow slugs of deleted items to be reused - they cannot be for workflows for instance. Because the slugs are not really handled uniformly though - pages do not have this problem and deleted pages can have their slugs reused because this is included in a check in the web controller which directly assigns valid slugs at that point. Slugs for deleted histories cannot be re-used - but they couldn't be prior to this commit - so this isn't making things worse in that respect.
In particular break out mixins defining context-based utilities for interacting with app, an user, and a history. In downstream work on workflow scheduling this proves nessecary and sufficient to create an context for 'execute'-ing tools (create jobs) outside of web threads in response to workflow requests.
Unit tests for some of this.
Dave Bouvier