Tom Lane wrote:
> A problem with such caching is that it'd fail to respond to changes in
> the content of pg_conversion. Now the code is already pretty
> insensitive in that respect, because if you're not doing any fresh "SET
> client_encoding" commands it won't ever notice changes in that catalog
> anyway. But this'd make it worse. We could ameliorate the issue
> somewhat by doing fresh lookups (and updating the cache) whenever doing
> SetClientEncoding with IsTransactionState() true, and only relying on
> the cache when IsTransactionState() is false.
>
> Comments?
Certainly seems like a reasonable compromise. From what I understand,
you'll get this "failed to commit..." message *if* you have changedf
things in pg_conversion. I think that's acceptable - it's not like
people modify pg_conversion all the time (at least I hope they don't).
//Magnus
