Istvan Toth
Istvan Toth
I don't think this is needed anymore.
Most of this doesn't even look relevant to memory usage. I seem to have mixed up some branches.
Can you take a look @julianhyde ? This is the patch discussed last week.
Sorry to bother you again, but can you make the time for this, @julianhyde ?
Can you open a [DISCUSS] thread on the calcite list about this change ? Explain its background in some detail ?
If we do this, then IMO the code cleanup and the check should be committed together. One hand this would be a major PITA, for backports, etc. On the other...
Tests fail because of PHOENIX-6493 Will not commit until after we've fixed PHOENIX-6493, and updated the default phoenix release in PQS to one with the fix.
A lot of the mentioned components are not in fact used in PQS.
I have opened https://github.com/julianhyde/sqlline/pull/478 for the same issue. That patch is smaller, and copies a solution already used in sqlline for numbers.
Can we revisit this @julianhyde ? I agree that for most DBs the current behaviour is fine, however it causes big problems with Phoenix, where some users have to downgrade...