Re: postgres 9.0 beta libpq empty binary array error

On 09/08/10 21:29, Heikki Linnakangas wrote:
> On 09/08/10 21:26, Tom Lane wrote:
>> Heikki Linnakangas<heikki(dot)linnakangas(at)enterprisedb(dot)com> writes:
>>> The behavior of empty arrays with dimensions is weird in general.
>>
>> Agreed, but we shouldn't be introducing random restrictions in the name
>> of security.
>>
>> Patch looks good to me, except that it occurs to me to wonder about
>> negative values of dim[i]. For small negative values this coding
>> will catch it, but what if it's large enough to overflow the other way?
>> Maybe use
>>
>> if (dim[i]< 0 || lBound[i]> ub)
>> ereport...
>
> ArrayGetNItems checks for dim[i] < 0. I concur though that it looks
> weird to not check that along with the overflow check, so maybe we
> should, just to make the code clearer.

Committed. I added a comment noting that dim[i] < 0 is checked in
ArrayGetNItems.

--
Heikki Linnakangas
EnterpriseDB http://www.enterprisedb.com