[CF-Devel] Status update

[Mark Wedel]

 As a note, I plan to make at least some performance improvements.  Hopefully
they don't break anything too critical.



>
     
      2) SOCKET. I have no proof that this is true. But the fact that
     
     >
     
         commands get updated as blocks leads to the suspection
     
     >
     
         that the recent changes to the socket-stuff might also
     
     >
     
         be involved here. I get this "disrupted character movement"
     
     >
     
         in rooms with lots of non-spellcasting monsters as well.
     
     >
     
         Of course to much lower extent, but still noticeable.
     
     >
     
         Or am I wrong and this is caused by something else?
     
     >
     
         Anyone got ideas?
     
     
 Not much has changed recently with the socket code that I am aware of.

 I will note that there could be several causes with the 'block of updates at
once':

 1) crossfire server is chewing up its time doing expensive processing, and when
it finally gets around to dealing with the player, it has lots of processing to
do, so it does a lot of player processing at once, hence a block of updates.
 2) Could be similar to above, but the problem is more that the socket buffers
as maintained by the OS are emptying, so it may not be doing much player
updating, but the block of data could be caused by the fact that the socket (OS
buffer) gets filled with a bunch of data it sends to the client.
 3) Could be something got broken, and the server is just sending a lot more
data that it did not send before or is not re-filling buffers when it should.

 It would be interesting to get some cpu statistic at the time of observed
behaviour.  IS cpu time maxed out (at least relative to number of cpu's, since
crossfire will only use one cpu)?  The cpu load is reasonable, that would point
to some issue with the socket code.  IF cpu is maxed out, then probably #1 or #2
is the cause, but the real fix is not to max out cpu time.