Possible bug - are dataset_preference rows being properly updated?

[jorvis]

When the user sets a display preference there might be a bug where a new row is being added rather the existing one being updated. This query should give a single row which is the owner's preference for a single-gene display on an example dataset, but it's returning three rows:

SELECT dd.label, dd.plotly_config
  FROM dataset d
       JOIN dataset_display dd ON d.id = dd.dataset_id
       JOIN dataset_preference dp ON d.id=dp.dataset_id
 WHERE dp.user_id = d.owner_id
   AND dd.user_id = d.owner_id
   AND dp.is_multigene = 0
   AND d.id = 'deb21a3b-677c-13e6-92cc-740fe8505e7c';

It's also worth noting that one of those is clearly a multi-gene configuration, so that also isn't being saved right somewhere (or at least at some point wasn't).

[adkinsrs]

We discussed this may be a "legacy" issue from an older dataset before code from save_default_display.cgi was fleshed out.

Relevant query from save_default_display.cgi:

        query = """
            INSERT INTO dataset_preference (user_id, dataset_id, display_id, is_multigene)
            VALUES (%s, %s, %s, %s)
            ON DUPLICATE KEY UPDATE display_id = VALUES(display_id);
        """