Oct 21, 2026, 6:16 PM@cpsdqs I'm not talking complexity bounds as a function of some N, I'm talking specifically computing a numerical bound for the given template-system, and storing that as input to inform execution or validation.
e.g. pre-allocating all necessary memory at start of runtime, not validating when checking, if the number is above a threshold, timeouts that error as a fatal "report this upstream" bug in the tooling if hit.
Oct 21, 2026, 6:21 PM@gaditb ohh huh. so basically turning it into an embedded programming sort of thing… that could be neat. …“(s?)sg that compiles to a program with exactly one malloc” lol
Oct 21, 2026, 6:23 PM@cpsdqs With exactly one malloc and... forward jumps only? Not sure what the computational restriction is to ensure measured, bounded, time as well.
Oct 21, 2026, 6:36 PM@gaditb <del>compile it to glsl</del> forward jumps only would make the program really big so maybe it’s okay to trust the system on that one. and have a pre-emptive multitasking os
Oct 21, 2026, 6:43 PM@cpsdqs What are GLSL (or its target)'s restrictions?
Oct 21, 2026, 6:49 PM@gaditb essentially, you can only use for-loops with a statically-ish known upper bound. combined with the statically known memory size and storage buffers, it would be possible to compile this as a massively parallel gpu-based website template renderer. now, whether this makes any sense,
Oct 21, 2026, 6:40 PM@cpsdqs EDIT:
Maybe "one single jump target at a time (which must be the active instruction when set), with a monotonically-decreasing-without-rollover jump-if-non-zero counter, which can't be forward-jumped over"? (EDIT2: forward-jumps clear the target&counter? no, 'cause we want IFs to be able to conditionally drop the counter /further/... hm.)
Then max time is naively "number of back-jump targets"*"max length before jump"*"max counter of jumps",
but can be more narrowly SUM("length from target to jump"*"counter set at target" OVER "back-jumps")