Re: [HACKERS] proposal: Support Unicode host variable in ECPG

On Thu, Oct 26, 2017 at 12:29 PM, Jing Wang <jingwangian(at)gmail(dot)com> wrote:
> I would like to provide the patch file that ECPG can support the utf16/utf32
> unicode host variable type. This feature is used to compatible the same
> feature in Oracle's. Please find Oracle's feature in the [1].
>
> [1] https://docs.oracle.com/database/121/LNPCC/pc_05adv.htm#LNPCC3273

The core of the patch is roughly 800 lines, and it adds 20k lines of
tests, which is large to digest.

+PQGetEncodingFromPGres 173
+PQGetEncodingFromPGconn 174
+PQGetEncodingName 175
+PQGenEncodingMaxLen 176
This adds four undocumented APIs to libpq. And why not using client_encoding?

The patch does not apply anymore and needs a rebase. There are also a
bunch of whitespace errors, no documentation in the patch, and it
seems to me that this patch introduces many concepts so it could be
broken down into many individual steps. Finally, I strongly recommend
that you review other's patches. You have two patches, including this
one, registered into this commit fest but your name is showing nowhere
as a reviewer. The patch you are proposing here is large, the usual
recommendation being to review one patch of equal difficulty for each
patch sent to keep the inflow and outflow of patches balanced.

I am adding Michael Meskes to this thread, he is the maintainer of
ECPG so perhaps he would be interested in what you have here.

Thanks.
--
Michael