ID:2674375
 
This week I finally got out 514.1553, and of course there was a new compiler bug that somehow evaded all the testing. I can't figure out how that sucker got through, but I did get it fixed for the release of 514.1554 yesterday.

SendMaps profiling has already been turning up some interesting data, and it should get more as more of the SS13 servers use it. I hooked it up to world.Profile() too for better support, although there's a bug in the JSON code in 1554. (Doh!)

What's most interesting so far is that I had a theory about where the biggest inefficiencies would be, and the first limited data that came in suggested the problem was in a completely different spot. So I started changing that code which did have a number of problems, only to get new data confirming my original theory. And SS13 is beginning to turn up data that's different from the previous cases, suggesting a possible third spot to look at.

Suffice it to say over the next week I plan to make big changes to the map-sending code to address all this, and hopefully greatly improve CPU usage. However the data that's available now is already helping projects even without engine improvements; Eternia managed to lop off more than half of its map_cpu time in response to the new metrics.

All of this work is made possible by all the Members and donors who contribute to keep development going, and I thank all of you who've been a part of that.

Hopefully the last of the snow is finally behind us and real spring is on the way. But until then, stay inside and keep building and playing those games! And keep collecting that new data, because I'm gonna put it to good use.