_dave_ wrote:I don't understand the point of the lowercase conversions, It seems entirely superfluous.
Yes, postgresql columns are not case sensitive. but lowercasing all the words does not solve the issue where you have both "cnt_hands_SB" and "cnt_hands_sb" as distinct PT4 columns? lowercasing them both simply makes them have a colliding name in PT4 also, no? So something else must be done, e.g. adding a suffix to one of the names. which on it's own would have prevented postgresql name conflicts and avoided this unfortunate scenario of breaking everyone's work.
You are right, but it would force column renaming instead because we are no longer lowercasing column names, we perform this duplication testing internally to detect potential cache column collisions instead. In 4.11.2 _SB is renamed to _SB_duplicate and a message is presented to the user stating that it needs renamed.
(to be more precise I am not positive is _sb or _SB would be renamed, unsure if its first come first serve or not but that should not matter for our explanation, its really just a behind the scenes issue).