At present, instructions that write to classical registers are immediately redirected to output, flushing any other instruction sequences that share that classical resource. This is unnecessarily aggressive: it is definitely possible (and even desirable) to perform compression on classical instructions, and it would be better to integrate these into the existing queueing logic. The main obstacle is extending the indexing of the queueing system to accommodate this, as it is currently tightly bound to the hardware-object instances that embody components on the quantum device itself.
At present, instructions that write to classical registers are immediately redirected to output, flushing any other instruction sequences that share that classical resource. This is unnecessarily aggressive: it is definitely possible (and even desirable) to perform compression on classical instructions, and it would be better to integrate these into the existing queueing logic. The main obstacle is extending the indexing of the queueing system to accommodate this, as it is currently tightly bound to the
hardware-object
instances that embody components on the quantum device itself.