Please don’t be patronizing insulting. What did read. What used to be one fairly quick query will now be thousands of decoupled queries. I know a number of cases where the decoupled query count can get into the hundreds of thousands. So what used to take 1 query and <2 minutes is now going to hammer the database with ~300k queries and additional logic on the programming side.

On Mon, Jun 3, 2019 at 4:50 PM Amir Sarabadani <ladsgroup@gmail.com> wrote:
I guarantee you, mediawiki core is everything but "over-normalized". We haven't done anything yet.

In WMCS it's slower for reasons mentioned above, in production it's fast. Also, as I mentioned about "join decomposition", please read.

Best regards

On Mon, Jun 3, 2019, 21:58 John <phoenixoverride@gmail.com> wrote:
This is a prime example of over normalization. We are normalizing and making things slower.

On Mon, Jun 3, 2019 at 3:38 PM Jesse Plamondon-Willard <pathoschild@gmail.com> wrote:
Sure! I just fetch the comment ID in the original query ([1]), and fetch the comment text with a separate query ([2]). You need n+1 queries based on the number of records, but the performance is adequate for my use cases.


On Mon, 3 Jun 2019 at 14:51, Huji Lee <huji.huji@gmail.com> wrote:
Out of curiosity, Jesse, can you share the query after that change, so it is all stored in this thread for future reference?
_______________________________________________
Wikimedia Cloud Services mailing list
Cloud@lists.wikimedia.org (formerly labs-l@lists.wikimedia.org)
https://lists.wikimedia.org/mailman/listinfo/cloud
_______________________________________________
Wikimedia Cloud Services mailing list
Cloud@lists.wikimedia.org (formerly labs-l@lists.wikimedia.org)
https://lists.wikimedia.org/mailman/listinfo/cloud
_______________________________________________
Wikimedia Cloud Services mailing list
Cloud@lists.wikimedia.org (formerly labs-l@lists.wikimedia.org)
https://lists.wikimedia.org/mailman/listinfo/cloud