Re: Patch for Prevent pg_dump/pg_restore from being affected by statement_timeout

-----BEGIN PGP SIGNED MESSAGE-----
Hash: RIPEMD160


> Sorry if I missed it in the original thread, but what is the
> use case you have in mind?

A use case would be dumping a large table and wanting to
load it into the database, but wanting to stop the job if it
is still running an hour from now, when a maintenance window
is scheduled to start.

However, I think my objection is more philosophical, as we've
changed from having pg_dump make a SQL representation of part
or all of your database, into also having it force what it
thinks should be the right environment as well. Yes, timeout
can be a foot gun, but it's a foot gun that off by default,
and must be explicitly turned on. The fact that a setting that
kills long-running queries ends up killing long-running queries
via psql -f seems worth a documentation warning, not a change
in the textual representation of a database. I checked the
archives and have yet to find a single instance in -bugs of
anyone complaining about this. The closest I found was someone
having problems with psql and -c, but they were specifically
aware of the timeout and were trying (unseccessfully) to
disable it first. For the record, I have no problem with
disabling the timeout in both pg_dump and pg_restore.


- --
Greg Sabino Mullane greg@turnstep.com
PGP Key: 0x14964AC8 200804171250
http://biglumber.com/x/web?pk=2529DF6AB8F79407E94445B4BC9B906714964AC8
-----BEGIN PGP SIGNATURE-----

iEYEAREDAAYFAkgHf/sACgkQvJuQZxSWSsiXOwCggD1P/SgPwOO3gJdlXKP5bU3l
dWgAnRK5FNixLy8ajgkfI3Y/UpDyoeZB
=qaA5
-----END PGP SIGNATURE-----