Re: ICU integration

On Sat, Oct 1, 2016 at 3:30 AM, Peter Eisentraut
<peter(dot)eisentraut(at)2ndquadrant(dot)com> wrote:
> On 9/23/16 2:27 AM, Thomas Munro wrote:
>> The warning persists even after I reindex all affected tables (and
>> start a new backend), because you're only recording the collation at
>> pg_collation level and then only setting it at initdb time, so that
>> HINT isn't much use for clearing the warning. I think you'd need to
>> record a snapshot of the version used for each collation that was used
>> on *each index*, and update it whenever you CREATE INDEX or REINDEX
>> etc. There can of course be more than one collation and thus version
>> involved, if you have columns with different collations, so I guess
>> you'd need an column to hold an array of version snapshots whose order
>> corresponds to pg_index.indcollation's.
>
> Hmm, yeah, that will need more work. To be completely correct, I think,
> we'd also need to record the version in each expression node, so that
> check constraints of the form CHECK (x > 'abc') can be handled.

Hmm. That is quite a rabbit hole. In theory you need to recheck such
a constraint, but it's not at all clear when you should recheck and
what you should do about it if it fails. Similar for the future
PARTITION feature.

--
Thomas Munro
http://www.enterprisedb.com