Over-allocation access

Description

These functions are experimental and somewhat advanced. By experimental we mean their names might change and perhaps the syntax, argument names and types. So if you write a lot of code using them, you have been warned! They should work and be stable, though, so please report problems with them. alloc.col is just an alias to setalloccol. We recommend to use setalloccol (though alloc.col will continue to be supported) because the set* prefix in setalloccol makes it clear that its input argument is modified in-place.

Usage

truelength(x)
setalloccol(DT,
    n = getOption("datatable.alloccol"),        # default: 1024L
    verbose = getOption("datatable.verbose"))   # default: FALSE
alloc.col(DT,
    n = getOption("datatable.alloccol"),        # default: 1024L
    verbose = getOption("datatable.verbose"))   # default: FALSE
setallocrow(DT, n = -1L)

Arguments

Details

When adding columns by reference using :=, we could simply create a new column list vector (one longer) and memcpy over the old vector, with no copy of the column vectors themselves. That requires negligible use of space and time, and long ago we did just that. However, that copy of the list vector of column pointers only (but not the columns themselves), a shallow copy, resulted in inconsistent behaviour in some circumstances. Therefore, data.table over-allocates the list vector of column pointers so that columns can be added fully by reference, consistently.

When the allocated column pointer slots are used up, to add a new column data.table must reallocate that vector. If two or more variables are bound to the same data.table this shallow copy may or may not be desirable, but we don't think this will be a problem very often (more discussion may be required on data.table issue tracker). Setting options(datatable.verbose=TRUE) includes messages if and when a shallow copy is taken. To avoid shallow copies there are several options: use copy to make a deep copy first, use setalloccol to reallocate in advance, or, change the default allocation rule (perhaps in your .Rprofile); e.g., options(datatable.alloccol=10000L).

Please note: over-allocation of the column pointer vector is not for efficiency per se; it is so that := can add columns by reference without a shallow copy.

setallocrow prepares columns for fast row operations (delete or insert) by reference. (Note that 'insert' by reference is not yet implemented.) Before such operations, columns must be resizable: ALTREP columns are materialized and all columns are made resizable (this might trigger reallocation).

Value

truelength(x) returns the length of the vector allocated in memory. length(x) of those items are in use. Currently, it is just the list vector of column pointers that is over-allocated (i.e. truelength(DT)), not the column vectors themselves, which would in future allow fast row insert(). For tables loaded from disk however, truelength is 0, which is perhaps unexpected. data.table detects this state and over-allocates the loaded data.table when the next column addition occurs. All other operations on data.table (such as fast grouping and joins) do not need truelength.

setalloccol reallocates DT by reference. This may be useful for efficiency if you know you are about to going to add a lot of columns in a loop. It also returns the new DT, for convenience in compound queries.

setallocrow modifies DT by reference to ensure all columns are resizable. Note that unlike typical by-reference operations, the underlying memory of each column vector is likely to be reallocated and relocated, if necessary. This means the memory addresses of the column vectors themselves may change, even though the data.table object DT is modified in place.

See Also

copy

Examples

DT = data.table(a=1:3,b=4:6)
length(DT)                 # 2 column pointer slots used
truelength(DT)             # 1026 column pointer slots allocated
setalloccol(DT, 2048)
length(DT)                 # 2 used
truelength(DT)             # 2050 allocated, 2048 free
DT[,c:=7L]                 # add new column by assigning to spare slot
truelength(DT)-length(DT)  # 2047 slots spare