Advent of Code via SnapLogic IIP

Day 11 is one of those days/puzzles where SnapLogic simply makes things take a long time, but there were some shortcuts that I could and did take to speed it up. Every time you have to preserve state, you effectively incur a time penalty (this is why the script snap can be so handy in situations like these). My current solution for part 2 still hasn’t completed, but I’m confident (when comparing to my code-based one) that it will work without any issues. Currently, part 2 (which requires 10,000 iterations) is running at 1.2 rounds per second, so that SHOULD finish in 2 hours, 19 minutes (roughly) which seems to be yet another repeat of Day 9 with respect to speediness. The key with speeding up these iterations, again, is using plex-local files (file:///tmp/file).

For this solution, I have 5 total pipelines, one for the full day (day11), then 1 for each major part (part 1 and part 2) as I found some performance impact by using a pipeline parameter to determine further downstream operations and flip the inner per-monkey processing loop. While the two inner processing loops are similar, they weren’t similar enough to really be able to use a pipeline parameter (in my opinion) to deal with it, though I did work with the example (4 monkeys) vs. input (8 monkeys) input in a fairly good way (in my opinion). You’ll notice that my “round” pipelines have 8 pipeline executes in a row, which all layer on top of each other, since I know at most we’d have 8 monkeys to process, I figured repeated pipeline executes made more sense in the end than trying to do another loop with saving state. I already knew that 10,000 iterations would be rough to do saving state for each of those, but multiplying that by 8 would slow it down even more, so in this case, taking a manual step can really help (if you know the scope of limitations. This is also why there’s a router and union for instances where the inner array doesn’t match the number being processed, it short-circuits the processing a bit more.

At midnight when I finished my coded solution, I never thought I’d be able to finish this puzzle using SnapLogic, and it’s a LONG runtime, but I’m still pretty impressed with being able to write up the solution and execute it in some amount of time that might be considered “valid” in some definition. You’ll notice the part 1 and part 2 specific items LOOK the same, and they’re both very close, but it’s the inner pieces that are different, I probably could combine those into just 2 pipelines rather than 4, but figured keeping down on the potential introduction of more issues was the best way to go forward, I’ll make a few more updates (after it runs successfully) and report back as necessary, but at least the outer loop should be able to be generic to both parts without too many issues.

Screenshots:

SLP Files:
day11_2022_12_11.slp (55.5 KB)
day11_round_part1_2022_12_11.slp (28.3 KB)
day11_process_monkey_part1_2022_12_11.slp (31.9 KB)
day11_round_part2_2022_12_11.slp (28.5 KB)
day11_process_monkey_part2_2022_12_11.slp (32.3 KB)