Sockets

[agabriel at home.tzo.org (Anthony Gabrielson)]

Do you have any doog ones I should look at?

Thanks,
Anthony

On Sun, 17 Oct 2004 dsr at tao.merseine.nu wrote:

> On Sun, Oct 17, 2004 at 08:52:32AM -0400, Jerry Feldman wrote:
> > 
> > There is a protocol called ASN.1 (http://asn1.elibel.tm.fr/en/) that is
> > used in some places. Essentially, everything is encoded into a TLD
> > (type-length-data) scheme where every data type is encoded as a byte
> > (eg. 32 bit int might be 1, char string (eg. octet) might be 2, user
> > defined 4, ...). The length is usually encoded in 7 bits. If the length
> > is longer than 127, then the length byte becomes a negative length of
> > the length. Example, sending a packet containing a string "abc", and int
> > 123:
> > 40X0A23abc13123
> > The 0x0A is the length of the packet 
> 
> Argh!
> 
> Bad ASN.1 parsers have been responsible for zillions of security
> holes in the last few years. 
> 
> When designing a protocol, please keep in mind the following
> concepts:
> 
> - first, if at all feasible, use an existing protocol. Best of
>   all a widely-known standard.
> 
> - second, if possible, extend an existing protocol. Make sure
>   you increment the version number or otherwise indicate your
>   incompatibilities with the original.
> 
> - third, design your protocol to look like a proven existing
>   standard protocol. Keep it as simple as possible.
> 
> - fourth, make sure it is at least human readable and writable
>   for the simpler exchanges. This will be of enormous help in
>   debugging.
> 
> - fifth, if you can't do any of the above, at least document the
>   protocol so rigorously that any half-competent beginning
>   programmer can write a working client. If you don't know
>   BNF, learn it.
> 
> -dsr- tired of gratuitously incompatible protocols
> _______________________________________________
> Discuss mailing list
> Discuss at blu.org
> http://www.blu.org/mailman/listinfo/discuss
>