It would be possible to improve client layer performance and probably squash any hanging bug that has crept in by doing some fairly major rearchitecting aimed at moving the cooldown queue to be entirely in RAM (so we don't actually unregister requests when they go into cooldown, we just remember not to send them for a while), and making it hierarchical (so we can exclude a whole section of the tree, e.g. a high level request rather than a segment, at once, for a period depending on when the requests under it will come out of cooldown).
This will involve significant code changes however, and may introduce its own set of bugs. What priority to give it? -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 197 bytes Desc: This is a digitally signed message part. URL: <https://emu.freenetproject.org/pipermail/devl/attachments/20100802/77206114/attachment.pgp>
