Re: Schema checking did not allow optional attributes (ITS#3889)

[ando@sys-net.it]

> Pierangelo Masarati wrote:
>
>>The suggested change looks incorrect; in fact, it must never occur that
>> the
>>a_vals member is non-null but no value is present.  Actually, it only
>> occurred
>>in the case indicated by (ITS#3861), so that issue could be correctly
>> tracked
>>down.  This change should be rejected.
>>
>>
> In actual testing, I've run into the case were a_vals is non-null but
> contains 0 items in the case of an optional attribute.  Perhaps a change
> upstream of the point in the patch would be more appropriate.

Sure.  Please feel free to report that problem thru the ITS, with adequate
information to track or reproduce it.  Optional attributes should not be
present at all; if an attribute is present in an entry, it __must__ have
at least one value, so that if "a_vals != NULL" then "a_vals[0].bv_val"
cannot be NULL.

p.

-- 
Pierangelo Masarati
mailto:pierangelo.masarati@sys-net.it


    SysNet - via Dossi,8 27100 Pavia Tel: +390382573859 Fax: +390382476497