Re: timeout implementation issues - Mailing list pgsql-hackers

From Zeugswetter Andreas SB SD
Subject Re: timeout implementation issues
Date
Msg-id 46C15C39FEB2C44BA555E356FBCD6FA4961D97@m0114.s-mxs.net
Whole thread Raw
In response to timeout implementation issues  (Jessica Perry Hekman <jphekman@dynamicdiagrams.com>)
List pgsql-hackers
> So the work that would need to be done is asking the driver to request the
> timeout via "BEGIN WORK TIMEOUT 5"; getting the backend to parse that
> request and set the alarm on each query in that transaction; getting the

Well imho that interpretation would be completely unobvious.
My first guess would have been, that with this syntax the whole transaction
must commit or rollback within 5 seconds.

Thus I think we only need statement_timeout. ODBC, same as JDBC wants it at the
statement handle level. ODBC also provides for a default that applies to all
statement handles of this connection (They call the statement attr QUERY_TIMEOUT,
so imho there is room for interpretation whether it applies to selects only, which
I would find absurd).

Andreas


pgsql-hackers by date:

Previous
From: Robert Schrem
Date:
Subject: Re: timeout implementation issues, lock timeouts
Next
From: Hiroshi Inoue
Date:
Subject: Re: Proposed patch for ODBC driver w/ C-a-n-c-e-l