På torsdag 16. juni 2016 kl. 20:19:44, skrev Tom Lane <t...@sss.pgh.pa.us <mailto:t...@sss.pgh.pa.us>>: Amit Kapila <amit.kap...@enterprisedb.com> writes: > On Mon, Jun 13, 2016 at 10:36 PM, Tom Lane <t...@sss.pgh.pa.us> wrote: >> min_parallel_relation_size, or min_parallelizable_relation_size, or >> something like that?
> You are right that such a variable will make it simpler to write tests for > parallel query. I have implemented such a guc and choose to keep the name > as min_parallel_relation_size. Pushed with minor adjustments. My first experiments with this say that we should have done this long ago: https://www.postgresql.org/message-id/22782.1466100...@sss.pgh.pa.us > One thing to note is that in function > create_plain_partial_paths(), curently it is using PG_INT32_MAX/3 for > parallel_threshold to check for overflow, I have changed it to INT_MAX/3 so > as to be consistent with guc.c. I am not sure if it is advisable to use > PG_INT32_MAX in guc.c as other similar parameters use INT_MAX. I agree that using INT_MAX is more consistent with the code elsewhere in guc.c, and more correct given that we declare the variable in question as int not int32. But you need to include <limits.h> to use INT_MAX ... regards, tom lane As of 4c56f3269a84a81461cc53941e0eee02fc920ab6 I'm still getting it in one of my queries: ORDER/GROUP BY expression not found in targetlist Am I missing something? I could dig into this further to reproduce if necessary. -- Andreas Joseph Krogh CTO / Partner - Visena AS Mobile: +47 909 56 963 andr...@visena.com <mailto:andr...@visena.com> www.visena.com <https://www.visena.com> <https://www.visena.com>
