limit results to contacts with out district info entered already.

[jmcclelland]

No description provided.

[MegaphoneJon]

Hi Jamie,

I think this is a move in the right direction, but two questions:

• Why do we group by contact_id here? In fact, I'm not sure why we're grouping by ID at all - I don't see how this query could return duplicates. That may be something I ported over from the SQL without looking closely.
• I agree with adding the new addWhere clause - but should we be removing the old one? It seems like this will cause us to redistrict addresses that have errors.

[jmcclelland]

I changed the group by to contact_id to be extra sure we don't return the same contact twice. I haven't tested but... what if we choose "Home" for the location type and a contact has two home addresses? Without group by, I think we would get two records from them which would eat two credits, but the second lookup would over write the first.

I also think we need to keep the old where clause. If we try a lookup and it fails, then electoral_districts.electoral_level will be NULL since nothing was retrieved and, hopeully, electoral_status.electoral_status_error_code will have a value indicating an error. By keeping that where clause, we ensure that we don't keep trying to re-district the same failing address over and over again.

[MegaphoneJon]

My main concern about the group_by is that it will give us an indeterminate electoral status error code in the scenario you describe - but multiple addresses of the same location type aren't really supported in Civi. You can create them with API or import, but then when you re-save one, it overwrites the other. So I guess I'm fine with this since both scenarios can lead to unexpected results, neither is supported, and neither is particularly worse than the other.