[bf1942] Map change process leak with 1.6rc2
AthlonRob at axpr.net
Sun Feb 8 21:11:10 EST 2004
On Sun, 2004-02-08 at 08:13, Steven Hartland wrote:
> Unfortunately is not a reliable method ( checking on map change may
> still detect a failure ), adds load even though small but most importantly
> is not consistent with not only the linux server but every other game out
> there so not an option :(
It seems to me it would be more reliable to check and see if the server
has a PID than to see if it has given you the console back (which, may I
remind you, doesn't work at all right now).
Without a PID, the server isn't running (and even when it is loading a
map, it has a PID). No PID = No server. No server = No PID. Seems
pretty straightforward and reliable to me. I suppose you don't really
need the sleep in there, either.
As for load increases... I would imagine the load increase would be
about on par with having a serial port driver loaded in the background.
You did say you were running Windows, right? So you can't be too
concerned about load...
The only difference between this and your current method is your method
assumes the server dies when it releases the console. This method waits
for the server to release the console, then it checks to see if the
server is still running. The only possibility of this breaking is if
the server doesn't hold the console while it is running; in this case
the loop would run and run and run, launching a new server every time,
which would be bad. But does that ever happen? Is that risk worse than
not having a working setup at all?
More information about the Bf1942