I find that i often have to keep multiple copies of my data files with different sortings, splittings into blocks and datasets, etc.
This uses quite some space, and tends to mess up my hard drive, apart from eating a lot of time to do the sorting.
I therefore propose a load/sort command for gnuplot, that creates the dataset to be plotted online. Possible syntax could be
sort "filename" [index ..] [every ..] [using ..] by n,m:o:p
where the sorting is first done over column n, each new n starting a new dataset,
then col(m), each new m starting a new dataset,
then col(o), each new o starting a new block,
and finally column p.
Minus sign before column numbers means inverse sorting.
Advantage to using an external program, creating temporary files, would be that i can already do some math before the sorting, with the familiar "using" construction.