Alfred Reynolds wrote: > > For each > client the network stream should remain ordered so in theory the > responses for a particular client would be in order but it may be > possible that they get out of order.
Wouldn't it then have been way easier to simply include the name of the cvar in the callback? void (*pfnCvarValue)(const edict_t *pEnt, const char * cvar, const char *value); With the current callback this feature gets overly complicated to use IMHO. Now one has to not only check several cvars one by one, but has also to compensate for any hickups that could occur in the network. That means implementing timeouts and queues and as soon as one answer goes missing there is no way to ever be sure that any subsequent answer is for the last query (except for pausing unpermissively long). It could just as well be a packet that got delayed badly. Plus, plugin systems like Metamod amplify these problems because multiple requesters could be querying at the same time. Which means extra queueing and extra mayhem when a packet gets delayed or even lost. Coming to think of it, that little extra parameter (including the name of the cvar, too) wold make things for anyone WAY less complicated and WAY easier to use. QUICK! You can still change it now! Nobody is using it yet! :) No, really, please consider a quick change. You didn't even change the protocol, so if you move fast the situation could still be remedied. I think that would help prevent lots of future trouble and I'm sure it would be appreciated by many. (Sorry if I joined rather late in the game. Kinda busy right now. :)) Florian _______________________________________________ hlds_apps mailing list hlds_apps@list.valvesoftware.com http://list.valvesoftware.com/mailman/listinfo/hlds_apps
