We are exposing `synapse get -q` to programmatically download data in portals from view queries (here is an example if you click Download Options > Programmatic Options).
I was surprised that when I tested this, the manifest was hidden in my Synapse cache and not the working directory.
When I test this with `synapse get` on a file. There is no manifest downloaded.
I thought I remembered this command downloads a manifest every time, automatically. What is the expected behavior with the manifest?
I'm testing this feature in the context of https://sagebionetworks.jira.com/browse/PLFM-5477.
Synapse Client 2.0.0
Good point about not conflating “manifest”. I think SYNAPSE_TABLE_QUERY_[file_handle_id].csv is suitable.
This is resolved and will be included in the next RC release.
A SYNAPSE_TABLE_QUERY_[file_handle_id].csv file will be downloaded to specified downloadLocation (defaults to the current working directory in the command line client).
code looks good!
The choice to create a name distinct from manifest is to distinguish the functional differences between a manifest and table query. In discussing with “It is currently more than just a distinction in the name, the table query produces a csv, manifests are tsv files, the columns of a tsv are proscribed (must have a path and a parent column, provenance columns and some other columns have special meanings, other columns are loose annotations etc). Whereas the table query file can have special row columns that aren't relevant to a "manifest" etc and depending on what the user entered as the query could include a lot of columns that are not interpretable as a manifest. they're just separate formats.”
Tested release candidate for version 2.2.0 and can confirm SYNAPSE_TABLE_QUERY_###.csv is in the working directory as expected.