On 08/06/04:31/5, Joe Conway wrote:
I just reread your comment and thought about it more. I think most of
what I said in the last post holds, but in addition part of the problem
lies in the fact that we don't implement multidimensional arrays as
nested arrays, but rather as one monolithic
Tom Lane wrote:
Joe Conway [EMAIL PROTECTED] writes:
select '{{},{}}'::text[];
ERROR: malformed array literal: {{},{}}
Hm. This seems like it would be a legitimate representation of a 2x0
array. Why would you allow '{}' and reject this?
Well, '{}' is a special case representing an empty array.
Joe Conway [EMAIL PROTECTED] writes:
Tom Lane wrote:
[ random griping ]
Well, '{}' is a special case representing an empty array.
Hmm ... good point ... it is not obvious whether that is an empty array
(0x0) or a 1x1 array containing a single zero-length string. I guess
if we want to make
Tom Lane wrote:
Joe Conway [EMAIL PROTECTED] writes:
select '{{},{}}'::text[];
ERROR: malformed array literal: {{},{}}
Hm. This seems like it would be a legitimate representation of a 2x0
array. Why would you allow '{}' and reject this?
I just reread your comment and thought about it more. I
Tom Lane wrote:
I'm not sure that we had such an agreement, but in any case we've got
to understand how to distinguish empty-string from NULL.
I realized that in my followup post regarding nested arrays. In any
case, since will not 8.0 support either NULL elements nor nested arrays,
can we agree
Tom Lane wrote:
I think that suppressing unquoted trailing whitespace is probably
reasonable. I'm much less enthusiastic about rejecting unquoted
embedded whitespace, though. I think that's significantly likely
to break apps, and it doesn't seem like it's really needed to have
sane behavior.
Joe Conway [EMAIL PROTECTED] writes:
The attached patch suppresses trailing whitespace, but allows embedded
whitespace in unquoted elements as discussed above. It also rejects some
previously accepted cases that were just too strange to be correct:
-- Postgres 8.0, with the patch
-- none
Joe Conway wrote:
I'm hesitant to apply the attached this late before the beta without
review, but it seems to take care of the pathological cases I came up
with, doesn't break anything AFAICS, and passes all regression tests. I
guess it can go into beta 2.
I've continued to hack on array
Joe Conway [EMAIL PROTECTED] writes:
... whitespace just before the delimiter is
significant (even though leading whitespace is not)
Yeah. This has been the documented behavior for quite some time, but
I can't say that I ever liked it.
I view the current behavior as a bug. While making