Sqlite 3.30.0 - Hackdot

Re:Sqlite 3.30.0 by asdkhadsj on Saturday October 05, 2019 @10:38AM (#21165755)

Recently I heard[1] that SQLite does not offer concurrent access safety. That is to say, the premise was that if you were to use it as a database backing something multi user, like a forum, SQLite would suffer due to the concurrent access of users against the DB.

While I've not had time to research it yet, I'd be curious on your (or anyones) knowledge / experience in this context.

My plan, if SQLite didn't offer concurrent access safety, was simply to put concurrent primitives around SQLite to ensure safety. However I'm not doing anything until I can make heads or tails of that original comment[1].

Thoughts?

[1]: https://news.ycombinator.com/item?id=21041433

[ Parent | on HN ]

Re:Sqlite 3.30.0 by thijsvandien on Saturday October 05, 2019 @12:44PM (#21166480)

I don’t have much experience with this, but I’ve been reading a bit about it last days. As far as I can tell:

- Concurrent reading has always been possible.

- Reading while writing is possible when the relatively new WAL mode is enabled, at no significant cost.

- Writing still results in a lock, and other writers can decide how long they’re willing to wait on it.

[ Parent | on HN ]

Re:Sqlite 3.30.0 by meddlepal on Saturday October 05, 2019 @11:08AM (#21165930)

Why not just devise a test rather than waiting for someone to hand you the answer? It's not a hard thing to simulate...

[ Parent | on HN ]

Re:Sqlite 3.30.0 by asdkhadsj on Saturday October 05, 2019 @01:15PM (#21166685)

I'm not expecting someone to hand me an answer. I was merely asking for experiences in this area. You think I'm going to rely merely on internet stranger to decide course of action on production systems?

God forbid I ask for experiences from people who seem seasoned in this area on Hacker News of all places. The gall I must have, right?

[ Parent | on HN ]