Investigate porting issue #57

[jezdez]

Give warning/error msg on inaccurate graph config

#57, PR: #91

Investigating since I'm not sure if there is a similar feature now upstream.

[alison985]

@jasonthomas Can you please re-enable my Athena connection so I can run tests cases on this?

[jasonthomas]

@alison985 sent you credentials via email.

[alison985]

I've been going back and forth with Jason. Asked my next question.

[alison985]

I can test this on my local with upstream master code now.

In PR #91 I implemented 2 warnings: warning_three_column_groupby and warning_three_column_stacking.

The original query with the problem in #57 has 2 dimensions and 1 measure. When I play with making charts based on that outcome I get an inaccurate chart if I don't have a groupby. The chart will work with or without stacking so I won't port that warning.

[alison985]

PR upstream getredash#3115

[jezdez]

Our PR has been made moot and this is most likely going to be fixed in getredash#1742 by adding the ability to aggregate same values.

I'm not sure if we should continue having this feature, since porting to redash-stmo is much harder given the code path involved. @emtwo @washort Do you have an opinion on this?

[washort]

As was pointed out in the upstream PR this does prohibit valid cases right now. I'd consider removing this and perhaps doing some discussion with users about the UX for creating these plots.

[jezdez]

We're closing this for now since we think we won't be porting this to upstream but will consider removing the feature once getredash#1742 is fixed differently. We may also just remove it once we get to a fork-free status since it's not a high-prio feature.