villebro commented on pull request #15320:
URL: https://github.com/apache/superset/pull/15320#issuecomment-868407355


   > @villebro would you mind re-reviewing this PR? I added all the additional 
time grains (not just week starting/ending) which have not previously been 
defined or the SQL connector. Let me know if this should be added in 
`db_engine_specs/base.py` (as defined) or whether `db_engine_specs/druid.py` 
should override the `get_time_grains()` method.
   
   I think this is fine. We have plans to introduce freeform timegrains (where 
easily supported), via regex or other pattern matching schemes. This would make 
it possible to do arbitrary time grains, like `PT7M`. Most engines support this 
fairly easily, but for others that are more restricted we would only support 
the built-in ones that have to be explicitly written out.
   
   > Furthermore I'm not sure if additional work is required, i.e., for Pandas 
post processing per 
[here](https://github.com/apache/superset/blob/7c17b1a97faae45883a141e5e70bfd4b66c902f0/superset/utils/pandas_postprocessing.py#L85-L102v).
 Additionally I'm not sure if the custom [time grain configuration 
logic](https://github.com/apache/superset/blob/master/superset/config.py#L544-L557)
 actually works if these need to be defined within the post processing as 
mentioned above, i.e., can Superset really support custom time grains?
   
   I think it's ok, at least for now, to just support the built-in time grains 
in the forecasting feature. But here we could probably fairly easily also 
support more custom increments via regex. I'd be happy to add this as soon as 
someone requests it or when we start working on the arbitrary precision time 
grain feature.
   


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

For queries about this service, please contact Infrastructure at:
[email protected]



---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to