You are viewing a single comment's thread from:

RE: Benchmarking hivemind API calls after switching to postgREST server

in HiveDevs8 days ago

Thank you very much for the outstanding work of your team. I believe that the deployment of hivemind based on postgREST will bring great improvements to the front-end site and various service programs.

In addition, I guess switching from Python code to SQL code should greatly improve maintainability, right?

Sort:  

Yes, it'll be easier to optimize the queries. Previously, performance issues could be in either the python or the SQL portion of the code. Now that it is all in the SQL code, we can just do explain on the queries and fully analyze the reasons for the performance.

switching from Python code to SQL code should greatly improve maintainability, right?

I think this works only if you know SQL at blocktrades level. For me, most of the SQL blackmagic in the current state is not easy-to-read :)

As far as I recall, a lot of the conversion of the mixed Python/SQL code was done by devs that certainly aren't SQL experts (although it was then reviewed by more experienced devs so that improvements could be suggested to the new SQL code), so either the code wasn't that difficult to understand, or at least the Python parts of it wasn't.

But in general, if you run into an SQL query that is difficult to understand (for sure some of them in hivemind are), I strongly recommend using an AI to comment the code for you, it does a good job.