Austin Bonander

This would close #1923 right? Unfortunately any change to null inference would be a breaking change so this will have to wait for 0.7.0. Fortunately we don't plan for the...

Proprietary DBs present a huge maintenance hazard for us as far as the open-source half of SQLx goes. I don't even see a free FOSS license offering on their site....

I'm wondering if there's a way that we can support incremental adoption of `cargo sqlx watch` into people's workflows. It is yet another process that needs to be running to...

@jplatte hmm, then as a compromise, maybe the macros could use the cached data _first_ if it exists and the migrations haven't changed. Then only the modified macro invocation will...

> As far as I can see, with `query_as!`, the rust types of output/input are known (parameters must be known because they are rust values, and output type is determined...

@loafofpiecrust have a look at this FAQ answer: https://github.com/launchbadge/sqlx/blob/master/FAQ.md#why-cant-sqlx-just-look-at-my-database-schemamigrations-and-parse-the-sql-itself It's a bit snarky but the suggestion comes up a lot and so it gets a bit annoying to respond to.

To determine nullability of columns in SQLite, SQLx will actually pull the generated bytecode for the prepared statement and step through it to find any expressions that could produce a...

@Bondza please open a new issue as that is a different database. Please include the output of `EXPLAIN (VERBOSE, FORMAT JSON) ` (or that of a similar query plus relevant...

cc @tyrelr who worked on the refactors to SQLite, sounds like some unintentional exponential blowup.

I went ahead and gathered a `perf` profile and it looks like an overwhelming amount of time is spent in `QueryPlanLogger::add_result()`, specifically `HashSet::insert()`:  And here's the output of `EXPLAIN`...