[tesrx]

>SQLite4 was an experimental rewrite of SQLite that was active from 2012 through 2014. All development work on SQLite4 has ended. Lessons learned from SQLite4 have been folded into the main SQLite3 product. SQLite4 was never released. There are no plans to revive it. You should be using SQLite3.

https://sqlite.org/src4/doc/trunk/www/index.wiki

[infogulch]

I love the complete clarity of this paragraph. There is no way to misinterpret it.

[eternityforest]

I'm kind of glad this didn't happen . It might have been nice if there was a performance boost, but it would have taken years for 4 to get to the level of trust 3 has, and possibly caused confusion for devs.

At this point sqlite3 is basically like UTF-8 or even ASCII. It will be there in basically any decent language. It's the default choice of embedded database.

Having 2 versions might de facto deprecate sqlite3 in some environments. There might be transition issues. And people would wonder if there will be a sqlite5.

The one thing I would have liked to see, is a sqlite-like database that operates on folders instead of files, and is optimized for being used in a modern multidevice environment, where files can't be efficiently modified, just overwritten, you don't have locking, you might want to version control things, and you want to reuse filenames if possible to avoid cluttering sync indexes with tombstones.

Other than that, sqlite3 seems to be good enough.

[SCLeo]

Thanks for the context. I totally thought SQLite4 is a new thing.

[nmstoker]

Yes, it's unfortunate the page doesn't indicate any useful dates for the content.

[beefman]

> Lessons learned from SQLite4 have been folded into the main SQLite3 product.

Did decimal arithmetic make it?

[cwillu]

Mostly no. There is an extension that provides a core of it (operators aren't overloaded, it's all function calls, and “all” is 3 operators total).