Thread: Re: Refactor ECPGconnect and allow IPv6 connection

On Fri, Mar 05, 2021 at 01:56:38AM +0000, kuroda.hayato@fujitsu.com wrote:
> I reviewed for myself and fixed something:
>
> * refactor parse_options(), same as conninfo_uri_parse_params() in libpq
>   Skipping blanks is needed in this functions because ecpg precompiler add additional blanks
>   between option parameters. I did not fix precompiler because of the compatibility.
>   If it changes, maybe SO_MAJOR_VERSION will be also changed.
> * update doc

As you are writing in your first bullet point and as mentioned
upthread, it does not strike me as a great idea to have a duplicate
logic doing the parsing of URIs, even if libpq accepts multiple
hosts/ports as an option.  Couldn't we have a better consolidation
here?
--
Michael

Dear Michael,

Thank you for replying!

> it does not strike me as a great idea to have a duplicate
> logic doing the parsing of URIs, even if libpq accepts multiple
> hosts/ports as an option.

Yeah, I agree your argument that duplicated parse function should be removed.

ECPG parses connection string because it uses PQconnectdbParams()
even if target is specified in the new-style,
hence my elementary idea is that the paring can be skipped if PQconnectdb() calls.

I will try to remake patches based on the idea.

Best Regards,
Hayato Kuroda
FUJITSU LIMITED