Re: BUG #5126: convert_to preventing index scan

Andrew Gierth <andrew(at)tao11(dot)riddles(dot)org(dot)uk> writes:
> "Tom" == Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> writes:
> Tom> Um ... locale *is* fixed within a given database, or at least
> Tom> LC_COLLATE and LC_CTYPE are. If you see cases where we have
> Tom> this wrong, they may need to be revisited.

> The value of LC_CTYPE etc. is fixed, but the meaning that the OS
> assigns to that value can be changed (arguably more easily than
> changing pg_conversion, now that we don't allow builtin conversion
> funcs to be used for conversions other than the one they are coded
> for).

[ shrug... ] So? We have no way of detecting that. By that logic,
float8pl isn't immutable because the hardware's FP behavior might
change under us.

My cutoff for this sort of issue is that if we provide a SQL command
to change something, we should expect that it will change. Otherwise,
it's the user's responsibility to deal with the consequences of a change.

regards, tom lane