In short: For as far as I can see, triggers can now only be built by checking items against static values (item <operator> <N>). It would be a tremendous advantage to be able to check against other items.

Example: now, it is impossible to add a decent trigger for high system load based on number of processes, since it depends on the size of the system. To be able to make a good statement on the load (other than looking at the load of the system), it would require a trigger statement like:

{template_linux:proc.num[,,run].last(0)}

>

{template_linux:cpu.numavailable.last(0)}

This would check if there are more runnable processes than there are CPUs installed in the system.

This could e.g. also be used to measure I/O load on, let's say a disk or controller against a theoretical maximum performance, which would give you an idea when the I/O is nearing the maximum possible.