Re: PATCH: Batch/pipelining support for libpq

On Fri, Nov 13, 2020 at 5:38 PM Alvaro Herrera <alvherre(at)alvh(dot)no-ip(dot)org>
wrote:

> Here's a v25.
>
> I made a few more changes to the docs per David's suggestions; I also
> reordered the sections quite a bit. It's now like this:
> * Batch Mode
> * Using Batch Mode
> * Issuing Queries
> * Processing Results
> * Error Handling
> * Interleaving Result Processing and Query Dispatch
> * Ending Batch Mode
> * Functions Associated with Batch Mode
> * When to Use Batching
>

Thanks!

I like the new flow and changes. I've attached a patch that covers some
missing commas and typos along with a few parts that made me pause.

The impact of memory came out of nowhere under the non-blocking mode
commentary. I took a stab at incorporating it more broadly.

The "are error conditions" might be a well-known phrasing to those using
libpq but that sentence reads odd to me. I did try to make the example
listing flow a bit better and added a needed comma.

Tried to clean up a few phrasings after that. The error handling part I'm
not especially happy with but I think it's closer and more useful than just
"emitted during error handling" - it gets emitted upon error, handling is a
client concern.

Seems odd to say the new feature was introduced in v14.0, the .0 seems ok
to imply. I didn't actually fix it in the attached but "the PostgreSQL 14
version of libpq" is going to become outdated quickly, better just to drop
it.

"The batch API was introduced in PostgreSQL 14, but clients can use batches
on servers supporting v3 of the extended query protocol, potentially going
as far back as version 7.4."

David J.