Global Hades issue tracker for Hades website

[ablack3]

Thanks to Paul Nagy we have all of the OHDSI github issues continuously loaded into an elastisearch instance with a number of attributes. We should be able to use the rest endpoints in the Hades website. As a first step I thought it would be nice to have a global Hades issue tracker that would allow sorting by various metrics (time open, etc).

[schuemie]

Great idea!

I hacked some JavaScript to show the current milestones, that might be an alternative to consider: https://github.com/OHDSI/Hades/blob/main/Rmd/milestones.js

[ablack3]

I need a service account to Paul Nagy’s massive GitHub database! Still waiting on that.

[schuemie]

What exactly do you mean by 'service account'?

[ablack3]

By service account I mean a username and password that is used by the app/website to access the elastisearch datastore containing all the mined github data. This would allow the website to query the datastore on the user’s behalf without requiring each user to have credentials.

[schuemie]

Sorry to be dense, but what is the elastisearch datastore? And who can create a service account for that?

[ablack3]

Not dense at at all! Happy to explain what I know. My understanding is probably not 100% correct though.

Elastisearch is an uberfast cloud data storage system for analytics. You can think of it like a database but it works differently that the usual SQL databases. It might be closer to a no-sql database or even a search engine.

Elastisearch has it's own query language implemented in JSON and an example query looks like this.

We can think of it like a database. We send a query and get results. The datastore I want access to (for the Hades website) is the one that powers this dashboard. The elastisearch datastore contains essentially all OHDSI github activity and is automatically updated. The fast performance of the elastisearch "database" would allow us to surface custom analytics about OHDSI repos on a public website. So for starters I was thinking we could have a table that would show all open issues and PRs sorted by various metrics (e.g. time since last comment). The idea would be to help us triage issues and PRs to tackle in a global/cross-repo way.

We could also just do this in the OHDSI kibana dashboard (link here) but I don't have permissions to create persistent changes here. If we had a service account to query the database we can build whatever we want on top of this fast query engine. I believe Paul (who set up the elastisearch datastore) is on board with this.

We could also use the github api directly but it would be much more cumbersome than querying this data in a superfast database.

[ablack3]

And who can create a service account for that?

@paulnagy I believe.

[schuemie]

Thanks! Now I think I'm up to speed.