Jack Christensen
Jack Christensen
In the native pgx interface you can entirely replace the logic any given type by registering a new Codec for that PostgreSQL OID. But I don't think you can change...
It's semi-intentional. The database/sql interface uses the type of the object being scanned into to determine how to parse the PostgreSQL data. All it knows is it got a `string`....
After a little further thought, I found a third place that might be better: `u?int(8|16|32|64)?Wrapper` (and I suppose `float(32|64)Wrapper`). If these implemented `TextScanner`, then they could try to parse any...
I just made a change (b593d5a4162fe368629bf71986ecf86bed88804d) for #2399 which may impact this slightly. I'm *fairly* comfortable with automatically parsing PostgreSQL text formatted values into Go numeric types even when the...
@kosmoz Can you check out #2183? @felix-roehrich has been doing some work on LoadTypes and it's possible his changes have fixed this.
It's not redundant. I just tried removing it and the tests hang.
Yeah, that's a tricky issue. The intention for context cancellation of a query was simply to cancel the query. If the query causes a connection attempt to be made that...
See #2185. I proposed something there that might have some overlap with this.
Do you have any benchmarks showing how much difference this would make? To be honest, I would be shocked if the length of the name of the prepared statement had...
> > To be honest, I would be shocked if the length of the name of the prepared statement had a measurable difference. As long as total message to PostgreSQL...