My current project is now completely non-blocking ...

# server

[bdawg.io]

My current project is now completely non-blocking behind coroutines EXCEPT for JDBC connecting to MySQL 😢

[nfrankel]

if the source is blocking, i suppose the rest being non-blocking offers no advantage
could you elaborate on this decision?

[bdawg.io]

In the case of JDBC, I have a thread pool for those queries to block on. So the calls are still asynchronous behind coroutines, but are still blocking on that thread pool.

I haven’t found a non-blocking replacement for MySQL database queries, so I’m stuck with JDBC at the moment.

[bdawg.io]

I would LOVE a truly non-blocking MySQL connection implementation though that I could put behind coroutines. The Ktor raw sockets look like a potential socket provider, but it’s not so simple to manually implement MySQL network traffic

[hallvard]

... and https://blogs.oracle.com/java/jdbc-next%3a-a-new-asynchronous-api-for-connecting-to-a-database ?

[hallvard]

oh, sorry, I saw this in the next thread only after posting ...

[bdawg.io]

Right, there’s these talks, but no real JRE support as of yet. There’s work towards it that appears to be in a “not ready yet…maybe next version” cycle for the time being

[bdawg.io]

Also, to note that ADBA is Oracle-specific the last that I have heard

It is not intended to be an extension to, or a replacement for, JDBC but, rather, an entirely separate API that provides completely nonblocking access to the same databases as JDBC.

[nfrankel]

thanks for your answer
did you benchmark before and after?
and if yes, can you provide the results?

[orangy]

Do you need a JDBC API, or just async execute?

[orangy]

Also, there is this: https://github.com/pgjdbc/pgsql2
It’s not MySQL, but a working SQL2 (new async DB API) implementation

[bdawg.io]

@orangy Just async access to MySQL is needed

[orangy]

“access” can mean anything 🙂 what operations do you need to execute? do you need some framework, like Hibernate to work with it? Other ORMs? Caches?

[orangy]

Do you need to read/modify a schema?

[orangy]

Or just send some raw SQL and get rows back?

[bdawg.io]

Nope, just the records themselves – SELECT, INSERT, UPDATE, DELETE
I use requery, but I’d be happy to work on getting around that

[orangy]

as far as I understand, #squash is very much like requery? I think both a based on ideas from #exposed

[orangy]

and jooq to some extent

[bdawg.io]

Doesn’t squash still have the same issue with JDBC being blocking?

[orangy]

Right now yes, but there is a branch with coroutines

[napperley]

Vert.x has its own async DB libs for MySQL/Postgres ( https://vertx.io/docs/vertx-mysql-postgresql-client/kotlin/ ) , MongoDB ( https://vertx.io/docs/vertx-mysql-postgresql-client/kotlin/ ), and Redis ( https://vertx.io/docs/vertx-redis-client/kotlin/ ).

[orangy]

@napperley yes, and we are building our own

[napperley]

Iiya - Will these libs be multi-platform?

[orangy]

Of course. But not very soon