[datatable-help] datatable-help Digest, Vol 28, Issue 2

[Matthew Dowle]

Ok, good idea. Will add that in to ?data.table ...

Thanks


> On 6/7/2012 12:46 PM, Juliet Hannah wrote:
>> Hi Yike,
>>
>> In Matthew's email, he writes:
>>
>> So in other words, when i has no key it's the first column of i to the
>> first column of x's key, the 2nd column of i to the 2nd column of x's
>> key,
>> etc.
> Right, I figured that was the part which was supposed to answer the
> question, however the language is ambiguous in Matt's reply, which is
> why I asked for clarification. If what you're saying is the case, then
> what the docs should say is something along the lines of...
>
> "If i is not keyed, then data.table performs a join on a positional
> match between the columns of i and the keys of X. In other words, a join
> is performed on (column 1 of i, column 1 of key of X, then column 2 of
> i, column 2 of key of X, etc...)"
>
> Don't have to use my wording, but the original was ambiguous to me.
>
> Cheers,
> Yike
>>
>>
>> I think this answers my original question. In the example (which is
>> from the FAQ), i has no key. So it seems
>> the  *default* is to match the first column of i to x's key.
>>
>> That's how data.table "knows". It uses the first column as default.
>>
>> At least that's the way I understand things now.  Matthew, please
>> correct if I've misunderstood.
>>
>> Regards,
>>
>> Juliet
>> i-bin/mailman/listinfo/datatable-help
>