Re: row_to_json bug with index only scans: empty keys!

On 11/07/2014 11:17 AM, Andrew Dunstan wrote:
>
> On 11/07/2014 10:51 AM, Ross Reedstrom wrote:
>> This is a serious bug in 9.3.5 and 9.4 beta3:
>>
>> row_to_json() yields empty strings for json keys if the data is
>> fulfilled by an index only scan.
>>
>> Example:
>>
>> testjson=# select count(*) from document_acl;
>>   count
>> -------
>>     426
>> (1 row)
>>
>> testjson=# SELECT row_to_json(combined_rows) FROM (
>> SELECT uuid, user_id AS uid, permission
>> FROM document_acl_text AS acl
>> WHERE uuid = '8f774048-8936-4d7f-aa38-1974c91bbef2'
>> ORDER BY user_id ASC, permission ASC
>> ) as combined_rows;
>>                               row_to_json
>> ---------------------------------------------------------------------
>> {"":"8f774048-8936-4d7f-aa38-1974c91bbef2","":"admin","":"publish"}
>
>
> That seems odd. Here's what the relevant code does:
>
>         td = DatumGetHeapTupleHeader(composite);
>
>         /* Extract rowtype info and find a tupdesc */
>         tupType = HeapTupleHeaderGetTypeId(td);
>         tupTypmod = HeapTupleHeaderGetTypMod(td);
>         tupdesc = lookup_rowtype_tupdesc(tupType, tupTypmod);
>
>    ...
>         for (i = 0; i < tupdesc->natts; i++)
>      ...
>
>             attname = NameStr(tupdesc->attrs[i]->attname);
>             escape_json(result, attname);
>
> Could this be a bug in lookup_rowtype_tupdesc()?
>
>
>


Further data point:

There's nothing json-specific about this, BTW:
   andrew=# select hstore(q) from (select * from idxo order by a) q;     hstore   ---------     ""=>"1"   (1 row)
   andrew=# set enable_seqscan = true;   SET   andrew=# select hstore(q) from (select * from idxo order by a) q;
       hstore   ------------------------------     "a"=>"1", "b"=>"b", "c"=>"c"   (1 row)
 


So it looks like the index scan only stuff is broken somewhere.

cheers

andrew