I'm replying through a webmail, so please bear with me. Quoting Kyle <[EMAIL PROTECTED]>: > 1. The Linux host runs fine, the Windows client crashes with an empty > random table. This is because it is unable to place its characters on > the map, and uses up all 1024 random numbers in the table trying.
Good catch! I did notice the 2 separately but hadn't linked the 2. > and sent. The other values that are placed in the table are never used > unless the server / client get out of sync. This seems like non-optimal > behavior. You mean, like adding some signature to validate that a random number is used in the correct place or something? > 3. Character placement fails because the client is always placing the > character in the ground. The only reasons I can see for this happening > is if either the random values are messed up or if what is considered > ground is messed up. When debugging this, it shows that it is trying > positions such as 20636071, 60589966. This works under linux, but I did find something suspicious (again) in Action. I can't attach a patch nor commit right now, so I'll just explain. Action::Push has a weird way to do the memcpy, using &tmp instead of tmp. The same was happening with PopDouble. I guess lodesi just copy'n'pasted code without noticing. I'm not sure how this impacts the network. > off from anything. One thing I have noticed is that RAND_MAX is defined > differently on Windows (MSVC) and Linux. This would affect a lot of > things in RandomSync I believe, and might explain some bugs. Might be worth checking what the definition is for mingw. Best regards, Kurosu _______________________________________________ Wormux-dev mailing list Wormux-dev@gna.org https://mail.gna.org/listinfo/wormux-dev
