iterview does not work for views on which you specify reduce in the options.

[mountainpaul]

So since your doing a reduce there is no "id" value, and it will throw an exception. It would be nice to fix this, but I think you would have to do a skip.

[djc]

Hmm, it feels like it might work with startkey, but without startkey_docid. Perhaps you can give that a whirl? The iterview() code could only pass startkey_docid if reduce is not specified, or even when id is not available on the resutl.

[mountainpaul]

Specifying a startkey does not appear to help. It appears for the first batch that 'id' is available, but on the second batch (i.e. next iteration) the id is not available.

[mountainpaul]

Changing client.py at line 870 from

           options.update(startkey=rows[-1]['key'], startkey_docid=rows[-1]['id'])
to 
           try:
                options.update(startkey=rows[-1]['key'], startkey_docid=rows[-1]['id'])
           except KeyError: 
                if rows[-1]['key']!=rows[-2]['key']:
                      options.update(startkey=rows[-1]['key'])
                else:
                      raise

fixed my issue, but in my specific case the key was unique because of the group level.

[nedgar]

When doing a reduce, isn't each key unique, even for different group levels? Seems like it should be feasible to just omit the startkey_docid when reduce is True. I'm trying this out now.