This is a "quick and dirty" hack. Do I like it? No. Do I want to delve
into further debugging/refactoring of [a 72 line method][1]? Not at the
moment. There are a few pathways forward, but for now, this is the
pathway to hopefully give us insight into the
**Why will this likely address the issue?**
Prior to this commit, for an experiment we would render each of the
goals (anywhere from 8 to 11); each of which would require 2 expensive
queries; which means about 16 expensive queries.
After this commit, the experiment page has none of those expensive
queries; instead each goal now runs the 2 expensive queries.
The hope is that this will help us show the results (albeit on multiple
pages) while coming in under the response time out handlers we have in
place.
**Why no tests?**
Ugh; I know right?!? I'm beginning to ask that myself. But for now,
because this is only visible to tech_admins, the consequences of
breaking are limited. In otherwords, this is not a customer facing
feature, so it can be a bit less robust in it's testing. At least
that's the rationalization I'm establishing.
Further, local tests would not reveal the production environment
complications of large data sets. The aforementioned expensive queries
are blisteringly fast on my local machine...in part because I don't much
field_test experiment data.
Closesforem/forem#17981
Related to:
- forem/forem#17895
- forem/forem#17869
[1]:ab2d7d29d0/lib/field_test/experiment.rb (L98-L160)
b8533172e3