Search Mailing List Archives


Limit search to: Subject & Body Subject Author
Sort by: Reverse Sort
Limit to: All This Week Last Week This Month Last Month
Select Date Range     through    

[bioontology-support] Update vs save an object in bioportal_web_ui

Paul R Alexander palexand at stanford.edu
Tue Apr 4 20:22:32 PDT 2017


I appears that the web_ui was sending the following parameters to update: {"views"=>["http://data.bioontology.org/ontologies/THE_VIEW"<http://data.bioontology.org/ontologies/THE_VIEW>], "splat"=>[], "captures"=>["MY_ONT"], "acronym"=>"MY_ONT"}

The front-end doesn't know about the inverse attribute, hence why it's sent. There may be a check on the back-end that is failing. I think inverse attributes should be discarded.

Another little note on the data model: why is the model crashing on the "attribute not loaded" problem? If the view attribute is asked but not loaded, then it could be automatically loaded. It would increase performance, and be way more comfortable for developpers: instead of having to use various methods to try to get all the attribute loaded (sometimes include, sometimes bring, sometimes goo_attrs_to_load, sometimes goo_aggregates_to_load) it would load dynamically. But I guess that now it would be a big change to do to the data model.

The decision to avoid loading data dynamically / automatically from goo-based objects was make developers explicitly aware that they were making a back-end call that could significantly impact performance.

Imagine a scenario where you are iterating over a list of 600+ ontologies and calling the views attribute on each, where each call to the attribute caused an HTTP request to be triggered. It would likely take a minute+ to complete the iteration in the best of conditions.

Best,
Paul



-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.stanford.edu/pipermail/bioontology-support/attachments/20170405/c1c17ab1/attachment-0001.html>


More information about the bioontology-support mailing list