Thread: pq_sendoob/pq_recvoob

pq_sendoob/pq_recvoob

From
Brett McCormick
Date:
Are these functions used at all?  A M-x tags-search didn't find them.
I'm not sure how they work over SSL (if at all).

Re: [HACKERS] pq_sendoob/pq_recvoob

From
Bruce Momjian
Date:
>
>
> Are these functions used at all?  A M-x tags-search didn't find them.
> I'm not sure how they work over SSL (if at all).
>
>

No, not used.  I think we thought they were passed unencrypted by SSL?

--
Bruce Momjian                          |  830 Blythe Avenue
maillist@candle.pha.pa.us              |  Drexel Hill, Pennsylvania 19026
  +  If your life is a hard drive,     |  (610) 353-9879(w)
  +  Christ can be your backup.        |  (610) 853-3000(h)

Re: [HACKERS] pq_sendoob/pq_recvoob

From
Tom Lane
Date:
Bruce Momjian <maillist@candle.pha.pa.us> writes:
>> Are these functions used at all?  A M-x tags-search didn't find them.
>> I'm not sure how they work over SSL (if at all).

> No, not used.  I think we thought they were passed unencrypted by SSL?

We are thinking of adding a "please cancel query in progress" function
to the FE/BE protocol, whereby the frontend could attempt to cancel a
previously issued query.  The cancel request would be sent from FE to BE
by an OOB message, so that the BE could detect it with a signal handler.
This would mean that cancellation would not work over an SSL link.
I'm willing to live with that, myself.

There are no cases where an OOB message is sent from BE to FE, and I
think we concluded that it would be too dangerous to try to do that.

            regards, tom lane