case sensitivity of jsonb keys and values

[kcranston]

From @bomeara on gitter:
"It now appears that looking up ot:focalCladeOTTTaxonName is now case sensitive. This gives studies:
curl -X POST https://api.opentreeoflife.org/v3/studies/find_studies -H "content-type:application/json" -d '{"property":"ot:focalCladeOTTTaxonName","value":"Mammalia","verbose":true}'
but this does not:
curl -X POST https://api.opentreeoflife.org/v3/studies/find_studies -H "content-type:application/json" -d '{"property":"ot:focalCladeOTTTaxonName","value":"mammalia","verbose":true}'
It came up b/c the R rotl package documentation has lower case examples that used to work but don't any longer -- it's easy enough to fix this in the documentation, but I just wanted to make sure this is intended behavior"

Originally opened in OpenTreeOfLife/otcetera#96 but really belongs here.

I think I understand why the case sensitivity, but I am confused about this being a change in behaviour (why did it ever work with lower case?)

[kcranston]

This search is an example of searching the jsonb data blob (i.e. ot:focalCladeOTTTaxonName is not an explicit column in the study table, but a key that might exist in the jsonb column for a study). Based on a github search, there are no instances of lower case 'mammalia' as ot:focalCladeOTTTaxonName in phylesystem. There is no explicit conversion to lowercase in the script that loads the nexsons, and dicts / json are case sensitive, as is the jsonb column in postgres.

I remain confused.

[mtholder]

Is it possible that:

1. an older version of the code (otindex, or peyotl, or phylesystem) did lower-case strings,
2. that older version of the code populated the running otindex db,
3. we changed something that causes the strings to not be lower cased,
4. the relevant studies were not been re-indexed with a newer version of the code that does not lower case things.
5. something triggered reindexing or recreation of the db, so now that rotl test fails

? could be tricky (and possibly not worth dissecting) if there was a long lag between step 3 and step 5 in that sequence