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
Submitted by: Matt Dowle; Assigned to: Nobody; R-Forge link
Export shallow().
DT2= shallow(DT)
would add reference counter attribute. (Both DT2[1,a:=5L] AND DT[1,a:=5L] would need to copy a. Maybe the ExternalPointer destructor could decrement as well, so a rm(DT2) would automatically take DT back to being the sole copy.
Then with this in place,
DT[, cols, with=FALSE]
# and DT[, list(col1,col2,...)]
could do shallow copies, too. See TO DO in source.
In the meantime, allow shallow() to accept a subset of columns and then use it in duplicated.R to save the copies of the by columns there. Upped priority for this part (easier).
The text was updated successfully, but these errors were encountered:
@mattdowle, the idea we recently discussed is already here (and written here in an even better fashion). Would be great to have this for next release! Also in my wish list for shallow() to accept a set of columns.. We could've shallow() and shallowv() like usual.
Submitted by: Matt Dowle; Assigned to: Nobody; R-Forge link
Export
shallow()
.would add reference counter attribute. (Both
DT2[1,a:=5L]
ANDDT[1,a:=5L]
would need to copy a. Maybe the ExternalPointer destructor could decrement as well, so arm(DT2)
would automatically takeDT
back to being the sole copy.Then with this in place,
could do shallow copies, too. See TO DO in source.
In the meantime, allow
shallow()
to accept a subset of columns and then use it induplicated.R
to save the copies of the by columns there. Upped priority for this part (easier).The text was updated successfully, but these errors were encountered: