On Mon, Nov 5, 2012 at 12:30 PM, Robert Muir <[email protected]> wrote:
> On Mon, Nov 5, 2012 at 12:23 PM, Michael McCandless
> <[email protected]> wrote:
>> On Mon, Nov 5, 2012 at 12:02 PM, Robert Muir <[email protected]> wrote:
>>> On Mon, Nov 5, 2012 at 11:47 AM, Michael McCandless
>>> <[email protected]> wrote:
>>>> Yes.
>>>>
>>>> But I think it's super-slow and not a real hang.   It's still running
>>>> for me but seems to be progressing ...
>>>>
>>>
>>> I'm just trying to figure out how this test coudl be so slow. it runs
>>> a single query against 3000 document index here :)
>>
>> It finally finished on my dev box: took 3,555 seconds.
>>
>> The problem is, for every hit, QueryUtils pulls a new scorer, and does
>> a skipTo/explain/etc. on that, and it does this like 4 times (if you
>> pass wrap=true).  This is especially costly for MTQs which do all the
>> up-front work to rewrite first... and throw in surrogate dance ...
>>
>> I think we should just pass wrap=false; I think it should make it 4X faster.
>>
>
> I don't think 1000 seconds would be that great either?
>
> The correctness of this stuff is basically tested in the other
> methods: the idea of this test was just to e.g. ensure deleted
> documents and that sorta thing worked.
>
> So i think we could make a little 5 or 10 doc index for this test
> instead of using the "big index" ?

OK I'll do that ...

Mike McCandless

http://blog.mikemccandless.com

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to