JanWieck@t-online.de (Jan Wieck) writes:
> ... So do we want to have indices storing plain
> values allways and limit them in the index-tuple size or not?
I think not: it will be seen as a robustness failure, even (or
especially) if it doesn't happen often. I can see the bug reports now:
"Hey! I tried to insert a long value in my field, and it didn't work!
I thought you'd fixed this bug?"
You make good arguments that we shouldn't be too concerned about the
speed of access to toasted index values, and I'm willing to accept
that point of view (at least till we have hard evidence about it).
But when I say "it should be bulletproof" I mean it should *work*,
without imposing arbitrary limits on the user. Arbitrary limits are
exactly what we are trying to eliminate.
regards, tom lane
