Ted Blackman

> as currently implemented it only checks the very latest snapshot. We may or may not want to check every single snapshot contained within the pier this way. The ship...

The reason to have num-live stored separately is just a cache so we don't have to do an N*log N traversal of the packet queue every time we need to...

Yes, someone would have to write code to identify and clear old-style Ames ducts to breached ships. It's low priority because this was a small leak that was not that...

I don't fully understand this issue or its fix. The fix seems to be for gall to pass a task to itself upon receiving a newly built agent core from...

This should be obviated by the agents-in-Clay work. — ~rovnys-ricfer https://tlon.io On Tue, Aug 23 2022 at 3:39 PM, Joe Bryan < ***@***.*** > wrote: > > > > >...

The `(met 3)` jet should just look at the size slot in the allocation cell, which says how many bytes the indirect atom has -- so, much faster than a...

If the first subscription update is over 100kB, then Gall will kick just after it. As long as the subscription is making progress (i.e. on resubscribe, the first message isn't...

Yeah, better not to ask devs to change anything for now.  Subscription reform will get rid of the idea of looping resubscription anyway. — ~rovnys-ricfer https://urbit.org On Sun, Jun 05,...

> We would be fine to say "at least 5 messages" here, right? In the hopes of not clogging earlier in any case. If it turns out this increases memory...

This seems like a problem with `%eth-watcher`. Maybe its timer is getting stuck. @dr-frmr How recent was your breach when this happened?