[ENet-discuss] Re: Stronger UDP checksum?

[Kevin Wasserman]

Certainly, the code is designed so that no crash should 

occur; the app just discards the data if it can't parse 

it.  However, it _does_ cause my lockstep application 

to go out of sync since the host believes it has 

successfully delivered a command (via a reliable packet) 

that the peer either doesn't execute; or, even worse, 

executes with inconsistent data.  That is "wreaking havok"

in my book.

 

-Kevin Wasserman

 

Bruce Mitchener wrote:

>Bad/corrupt data shouldn't crash an application or server though.
> 
>  - Bruce
> 
>Lee Salzman wrote:
>> Send me a patch for it, and I will make it a compile time option.
>> 
>> Kevin Wasserman wrote:
>>> I've got a user who seems to occasionally (every
>>> 5 minutes or so, which is frequently enough) transmit
>>> a packet that becomes corrupted but apparently
>>> passes the UDP checksum and then naturally
>>> wreaks havok on my app. I was wondering if
>>> anyone else had had a similar exerience and
>>> thinking that I might add an option to enet to
>>> provide a stronger checksum (e.g. CRC32) when
>>> sending a packet. For reliable packets, particularly,
>>> I think it makes sense to do it in enet rather than
>>> at a higher level. Any thoughts?

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.puremagic.com/pipermail/enet-discuss/attachments/20050610/70beb810/attachment.htm