Hannu Krosing <hannu@tm.ee> writes:
> Is there no simple (i.e. cheap) way to disallow \0 alltogether for
> these types then ?
> perhaps just strip them out in textin() (or is it text_in()) ?
They *are* stripped out in textin(), by virtue of the fact that
textin expects to see a null-terminated input string.
>> It wouldn't surprise me too much if there are inconsistent
>> behaviors between indexscans and seqscans for such invalid data.
> should'nt they both use the _same_ strcoll() and friends ?
Irrelevant; the issue is that the various comparison operators may
produce inconsistent results given invalid input. For instance
texteq() short-circuits to a FALSE result if the lengths of the
inputs are different, which means that 'ab\0' = 'ab' will produce
false, even though a strcoll-based comparison will claim they are
equal. I don't think that means that texteq() is wrong to check the
lengths first.
> Should we not examine "the _possible_ outputs of every C-coded function
> to make sure it produces a valid value of the datatype" ;)
Go for it.
Possibly chr() should reject chr(0) ...
regards, tom lane
