On Fri, May 20, 2016 at 2:02 PM, John R Pierce <pierce@hogranch.com> wrote:
> On 5/20/2016 10:56 AM, Kevin Grittner wrote:
>
> ... or use the --quote-all-identifiers option of 9.5 pg_dump.
> http://www.postgresql.org/docs/9.5/static/app-pgdump.html
>
> Not a bug.
>
>
> pg_dump and pg_dumpall are supposed to be able to dump all supported prio=
r
> versions, thats the documented procedure for updating... I contend thi=
s
> *is* a bug. quote-all-identifiers is a workaround, for sure.
>
>
> =E2=80=8BThis was my first reaction, and I suspect that we attempt to do =
this
already and that this one in just an oversight=E2=80=8B.
If indeed the prescribed procedure is quote-all-identifiers I'd say we
should at least update the docs.
"""
Force quoting of all identifiers. This may be useful when dumping a
database for migration to a future version that may have introduced
additional keywords.
"""
=E2=80=8BI'd probably say something like:
"It is recommended to use this option when performing dumps intended for
migration or when dealing with a version of PostgreSQL different than
pg_dump. This is because the recognized keywords sometimes change between
major versions but only the quoting rules of the pg_dump version are
applied."
This doesn't seem to make its way that often to these lists otherwise I'd
recommend we add a "no-force-quote-identifiers" option and default to this
behavior.
David J.