Groups with large inventories slow down pipe running

[fechan]

If you pipe to/from groups with large inventory sizes (even if most of the inventory is empty), the time it takes to process that pipe increases significantly. Are we requesting the item details of every slot even if they're empty, or is there something else going on?

[fechan]

The answer is yes: item details and slot limits are both being requested even if they're empty.

Some optimization ideas

• If the number of slots in the sources+destinations combined is greater than the number of parallel runners, then it's worth requesting the non-detailed info first, then fulfilling details afterwards for non-empty slots
• The slot item limits are unlikely to change after the first call. Rare cases where it can change include the Create adjustable barrel. I think it's worth having a setting on each group or machine that tells SIGILS to request item limits every pipe run, and have it off by default.
  • This would mean persisting the ItemDetailAndLimitCache over different pipe batches, so there needs to be a function to wipe the item detail cache but not the item limit cache.
• If we already have non-detailed info of both src and target, we could skip getting item details of any slot in the target group that doesn't match item ID with any stack in the source.