detule

@shrektan Thanks for writing up the documentation; I was able to follow along better. Nice work again. This is very minor, but I was wondering about this: Rather than introducing...

@smmurphy: * On the first issue (`to_bytes`) - can you verify which version of GCC you are using? `gcc --version` perhaps. * On the second (truncated columns) can you verify...

Thanks - I don't have access to those drivers but [looks like there are some config options](https://docs.rstudio.com/drivers/1.4/pdf/Simba%20ODBC%20Driver%20for%20PostgreSQL%20Install%20Guide.pdf) (`MaxVarchar`, perhaps) that might help with the truncation issue.

Hi - just looking at the NEWS file: one of the things that changed going from 1.2.3 -> 1.3 specific to insertion was the default value for the`batch_rows` argument. Can...

@krlmlr I am struggling to reproduce with freeTDS 1.2.11 - in that it seems to me DATETIME2 is coming back as POSIXct (maybe i misunderstood the issue). There I think...

We are off-topic here from the original issue, but for me both DATE and DATETIME2 come back appropriately class-ed with FreeTDS: ``` > values dbWriteTable(con, "test_date_time2", values, field.types = list(c1...

Hey Jim - Isn't SQLCancel called by nanodbc when some of the resources are destroyed? If so, I wonder if we can get there by looking into opportunities to `release`...

That makes sense; I think the `checkUserInterrupt` call that Jim referenced above is also at the point where the query is being submitted for execution, prior to any `fetch`ing. He...

Hey @hadley, @jimhester: I don't think anything we do here, short of real `async`, can make the query in that example interruptible, right ? (just making sure that's not the...

@hadley - No, I think the goal of wrapping the existing `checkUserInterrupt` calls (and/or introducing a few more) in try-catch blocks, would be to make sure that we reset the...