Thought some of you fools might want a little work-related update; from today’s beta build:
– Improved inter-host peering
– Improved inter/host connection loss and recovery
– Improved cell host chat host connection
– Added new performance counters and tracking
Host Internal Testing:
– On startup, strat host builds a collision matrix for enabling the very-rapid detection of which CP a player is near
– On startup, strat host builds a TOE-style supply/delivery matrix for performance/viability testing
Some of this is just fixing a couple of issues that I found that have been in our system since day 1. Hopefully some of this should usher in a new era of host stability – finally fixing one of the longest running stability bugs; sadly its mere coincidence that I knew where to look for it – testing something unrelated I happened to set a breakpoint on the very line causing the issue.
For the rest, we had to pull back from my prior TOE implementation when it proved to be a major resource conflict with the rest of the strat server. I tried various ways of working around it until finally we took a step back, reviewed the design and requirements, stripped away a few more “fluff” elements of TOEs, clarified some others and really came away with the feeling this was a design we feel good about.
I’ve taken a couple of days stewing these changes and rolling back my mental picture of the servers and leapt back into the fray getting a host with a skeletal TOE-engine in it that is instrumented for testing and evaluation via the Lua-based scripting testing truss I’ve cobbled together over the last 2 years.
That doesn’t mean its on display yet – infact beta testers should be unaware of it.
Part of this is the “collision matrix”. This is really a chance to avoid having to “port” part of the strat/supply/spawning system from non-TOEs to TOEs as well as another step towards replacing the “grid” system in the cell hosts: some of you may recall that the area chat “grid system” (which is also used by STOs) was a dry-run for using a different mechanism for our world grid. It’s still very rough and aside from sparing me from having to shovel horse-shit code into the TOE system, it doesn’t really have anything exciting to describe at this point.
On the client side: there’s some really nice stuff in this beta and this is the first time we’re using Trac to back-end our beta testing/confirmation/ticketing process so there’s a bit of a positive buzz in the office today.
Some of Sebastian (Jaeger55)’s work has finally been going into source control and they’ve been working through the week to get that all integrated – I don’t yet know if it made the beta, I’ll be checking it out from home when I get there. Killer is incredibly eager to see what we’ve seen so far make it into the game, though.