RE: pgsql: Add a new GUC and a reloption to enable inserts in parallel-mode

From: Robert Haas <robertmhaas(at)gmail(dot)com>
> The amount of code isn't the issue. I'd rather expend a little more
> code and solve the problem in a better way.

Reading your reply to Andres-san, I feel sympathy about your attitude. Maybe we should outline the (rough) design first, discuss/guess its complexity, and ask for opinions on whether it's worth expending our effort for the complexity or choose an easier hack, considering the assumed use case and ensuring future extensibility not to prevent smooth enhancements.

With that said, I think the easy hack this time is good, because parallel INSERT/UPDATE/DELETE will only be used infrequently (compared to short OLTP transactions) for data migration and batch processing where the source data is ample. I don't think of a use case yet where we want to make parallel workers allocate an XID when only necessary.

> I did not mean to give offense, but I also don't think grotty is a
> strong word. I consider it a pretty mild word.

Thank you for telling me the impression of the word. TBH, I knew the word for the first time, and looked it up in English-Japanese dictionary. It didn't give good impression, and grotty sounds somewhat similar to grotesque, so I got worried "Robert-san may be angry."

Regards
Takayuki Tsunakawa