You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
(and eventually to any type of logical operation). This would be very efficient as the number of condition increases (as binary search reduces the search space with every condition).
My PR #2494 addresses the optimized subsetting on multiple columns with & connection like DT[a==1 & b == 2]. Additionally, support for the %chin% operator was introduced. However, <, <=, >, >= are not yet supported. Neither are queries that are connected by | like DT[a==2 | b==3].
I am not sure if we should keep this issue open. Auto index has been extended to work on multiple columns by @MarkusBonsch PR. The OR operator is not yet there, but as per initial comment
as binary search reduces the search space with every condition
It was not really asking for | but for &.
Unless there is an idea to address other stuff, I would close the issue, it is quite broadly defined and biggest part of it has been already addressed. If we want to keep it open, it is best to list precisely what is to be done, so it will be clear what is left.
Activity
franknarf1 commentedon Feb 16, 2016
This came up on SO: http://stackoverflow.com/q/35438190/1191259 Might want to post an answer there when this is implemented.
Another (though it might get deleted): http://stackoverflow.com/q/43635968/
jangorecki commentedon Feb 16, 2016
without binary search it could be done now, if using under the hood with
Reduceoverx[i = ..., which = TRUE]many queries with indexes and combined.arunsrinivasan commentedon Mar 9, 2016
Moving this to next release (should be straightforward after non-equijoins are done (hopefully)).
MarkusBonsch commentedon Nov 22, 2017
My PR #2494 addresses the optimized subsetting on multiple columns with & connection like DT[a==1 & b == 2]. Additionally, support for the %chin% operator was introduced. However, <, <=, >, >= are not yet supported. Neither are queries that are connected by | like DT[a==2 | b==3].
jangorecki commentedon Apr 3, 2020
I am not sure if we should keep this issue open. Auto index has been extended to work on multiple columns by @MarkusBonsch PR. The OR operator is not yet there, but as per initial comment
It was not really asking for
|but for&.Unless there is an idea to address other stuff, I would close the issue, it is quite broadly defined and biggest part of it has been already addressed. If we want to keep it open, it is best to list precisely what is to be done, so it will be clear what is left.