Posts made by eth0nic

4*50/60 =  3,3 minutes to cross a single room is kind of unacceptable. What is the bottleneck? The cpu load? The mongoDB backend?

The whole game should really mostly scale linearly as there are little to no dependencies to care about between rooms, etc. ... so yes, it should scale horizontally quite well. The initial poster is more than right ... for 10Eur/month I can get a dedicated AWS instance and not even the smallest one ...

I just want 2sec ticks back ... badly.

*edit* others posted my <200 lines solution already ...

And now we are at 4.1 sec avg() ... Way too slow in order to really enjoy watching ... plus the Steam client wrapper thing is lagging so hard ... maybe we can have a dedicated client written in C++ or something ... not like a few creeps should be a problem to handle if you check out RimWorld or Factorio ...

So what's the command? Anybody found out in the last 7 month since the feature (allegedly) got implemented?

Good Lord!!!

Can not attach pictures in this section and editing the initial posting does not actually change it.

See picture here: http://support.screeps.com/hc/en-us/requests/2110

The secondary color part of some flags is not drawn properly. See picture below.

Thank you for taking the time to point this out. I forgot about the weird diagonal thing. So it basically gives you the shortest path just "biased" towards the beginning of the data structure. Makes sense but Pythagoras disapproves.

controller.pos={"x":9,"y":38,"roomName":"W27N82"}

exit to W26N82 / exitDirection= 3
roomExit= {"x":49,"y":5,"roomName":"W27N82"}

exit to W28N82 / exitDirection= 7
roomExit= {"x":0,"y":29,"roomName":"W27N82"}

Totally not the closest exit tiles ... see the blue lines:

findClosestByRange(exitDirection) returns a random tile of the correct exit but not the tile closestByRange to the invoking object.

My controller is at x9/y38, the returned closestByRange exit tile is x49/y5 while the exit literally spans the entire room and the closestByRange exit tile should be x49/y38.

Please fix - thank you!

A call to spawn.renew() returns "ERR_FULL" for creeps which are nowhere near 1500 ticksToLive.

.renewing a 1MOVE only creep (scout) results in the creep never going beyond 1270 ticksToLive because the .renew call would "over-heal" the creep it seems. Bigger creeps can't be .renewed beyond 1450-ish ...

Please fix the check which is preventing the .renew up to 1500 TTL - thank you.

This broke my code as well. This is code breaking, most likely caused by a simple typo like "userane" (as the client shows the "username" field is stored in mongo) and still hasn't been fixed?