search

chop-dbhi / cilantro

Official client for Harvest (http://harvest.research.chop.edu)
http://cilantro.harvest.io
Other
28 stars 8 forks source link

Query history #365

Open leipzig opened 10 years ago

leipzig commented 10 years ago

Would like query history, similar to explicitly saved or shared queries.

The rollover of the names need to be somewhat concatenations of the filter itself (Subject Race is Asian, Age is Between 0.0 and 2.0)

screen shot 2014-01-27 at 2 56 29 pm

naegelyd commented 10 years ago

@leipzig Can you provide some more information on what exactly is being requested here?

leipzig commented 10 years ago

Right now an old query is only available if a user explicitly saves it. Why not just keep a history of every query a user performs?

naegelyd commented 10 years ago

@bruth @mitalia Thoughts on this? I am in favor of keeping things as is. If we were to track every query the user performed, I would have thousands of queries just from my testing and I'm sure some of the Varify users would be in the same boat. I am not really sure I see the value in this. If a query is important for some reason, it can be saved. In my opinion, keeping track of all queries will quickly result in an unwieldy list of mostly noise.

I do think that part of the requested functionality could be useful. I think the query list could be updated so that the not only the name of the query is shown, but also, under each query, we show the filters that comprise that query. This way, even if a query is named poorly, the user should be able to tell if it is the one they want by looking at the filters without having to open the query just to ge the filter list.

bruth commented 10 years ago

We actually already do save every revision: https://github.com/cbmi/avocado/blob/master/avocado/models.py#L1039-L1044

leipzig commented 10 years ago

so in pubmed there is a "Recent Activity" panel that behaves this way. Incredibly useful there since virtually no one gets their pubmed queries right the first time screen shot 2014-05-07 at 1 50 53 pm

bruth commented 10 years ago

And have Serrano resources for them: https://github.com/cbmi/serrano/blob/master/serrano/resources/query.py#L346-L354

We just don't have a UI for it..

mitalia commented 10 years ago

I think the query history is a useful feature with one major caveat: In production we have often had metadata changes that have broken saved queries and session queries requiring manual migration. On a long-running project, a person's query history may have multiple queries that become invalid during a new release. We have historically been pretty bad about managing this situation, though @bruth and I long ago discussed approaches to at least rendering the newly invalidated query and giving the end-user the ability to modify it. I'm not sure how many of those ideas have trickled into the codebase over time. If that is all resolved and the UI can give the user tools to properly fix a now defunct query, I see no reason not to have this capability.

naegelyd commented 10 years ago

I retagged this as 3.0 so we can work on it in the future when there is time and the UI has been more thought out.