On Mon Jul 24, 2023 at 12:43 PM CDT, Tom Lane wrote:
> "Tristan Partin" <tristan@neon.tech> writes:
> > v3 is attached which fixes up some code comments I added which I hadn't
> > attached to the commit already, sigh.
>
> I don't care for this patch at all. You're bypassing the pqsignal
> abstraction layer that the rest of psql goes through, and the behavior
> you're implementing isn't very nice. People do not expect ^C to
> kill psql - it should just stop the \c attempt and leave you as you
> were.
>
> Admittedly, getting PQconnectdbParams to return control on SIGINT
> isn't too practical. But you could probably replace that with a loop
> around PQconnectPoll and test for CancelRequested in the loop.
That sounds like a much better solution. Attached you will find a v4
that implements your suggestion. Please let me know if there is
something that I missed. I can confirm that the patch works.
$ ./build/src/bin/psql/psql -h pg.neon.tech
NOTICE: Welcome to Neon!
Authenticate by visiting:
https://console.neon.tech/psql_session/xxx
NOTICE: Connecting to database.
psql (17devel, server 15.3)
SSL connection (protocol: TLSv1.3, cipher: TLS_AES_256_GCM_SHA384, compression: off)
Type "help" for help.
tristan957=> \c
NOTICE: Welcome to Neon!
Authenticate by visiting:
https://console.neon.tech/psql_session/yyy
^Cconnection to server at "pg.neon.tech" (3.18.6.96), port 5432 failed:
Previous connection kept
tristan957=>
--
Tristan Partin
Neon (https://neon.tech)