Sorry, upon re-reading I guess my initial post might have sounded naive.  We
already have a plan for a "custom validator" that can verify a result
without checking it against its neighbor.  I was really just wondering,
API-wise, how to do validation on a single result as opposed to two.  thanks
again

On Fri, Oct 30, 2009 at 2:13 PM, Adam Bazinet <[email protected]>wrote:

> Thanks, we're aware of the potential issues here :-)
>
> Now does anyone have an answer to the technical question?  I suppose this
> also applies to anyone using the AdaptiveReplication scheme (which we're
> not)
>
> thanks,
> Adam
>
>
>
>
> On Fri, Oct 30, 2009 at 2:02 PM, <[email protected]> wrote:
>
>> Unless you have a quick method of getting back from the answer to the
>> original question (i.e. one direction is hard and the other is extremely),
>> do yourself a favor and use 2 tasks for validation.  Some projects have a
>> means of validating the result against the original answer and others do
>> not.  It really depends on the problem space your project is in.
>>
>> jm7
>>
>>
>>
>>             Adam Bazinet
>>             <pknut...@umiacs.
>>             umd.edu>                                                   To
>>             Sent by:                  [email protected]
>>             <boinc_dev-bounce                                          cc
>>             [email protected]
>>             u>                                                    Subject
>>                                       [boinc_dev] single result
>>                                       validation
>>             10/30/2009 01:59
>>             PM
>>
>>
>>
>>
>>
>>
>>
>>
>> Hi,
>>
>> In the BOINC validator framework, what is the recommended way to validate
>> just one result?  The compare results function takes two results, of
>> course,
>> so I probably don't want to use that (right?)
>>
>> Is there some example code in the documentation or distribution somewhere?
>> Or would any projects doing single-result validation be willing to share a
>> template of their code with me?
>>
>> Currently we are using a default string comparison validator and the
>> two_credit function - we're starting with two results/WU, but would like
>> to
>> cut that down to one for a particular application.
>>
>> thanks,
>> Adam
>> _______________________________________________
>> boinc_dev mailing list
>> [email protected]
>> http://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev
>> To unsubscribe, visit the above URL and
>> (near bottom of page) enter your email address.
>>
>>
>>
>>
>
_______________________________________________
boinc_dev mailing list
[email protected]
http://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev
To unsubscribe, visit the above URL and
(near bottom of page) enter your email address.

Reply via email to