Re: NULL-Request (Sect. 4.1)

>This is not as good as saying it is at the beginning, from the
>standpoint of straightforward parsing. It is harder to look for optional
>stuff at the end, when no more stuff may be forthcoming -- I don't know
>how long to wait to see if it arrives. 

Paul here encapsulates the nub of the argument. This may be the way we may wish 
to present the problem. It isn't a solution for the server writers.

Server writers need to be aware of the bogus CRLF problem and "cope". This may 
lead to butt-ugly code but sending spurious data was a butt-ugly hack in the 
first place. There are TCP/IP stacks which break unless the server writer codes 
arround this lunacy. 

I think we all agree that we don't like the fact that the iceberg exists. The 
problem is to avoid it holing the ships :-) 



I think Jeff's wording with clients must not do the wrong thing, servers must be 
aware that this is a very common form of garbage is a good one.


	Phill

Received on Wednesday, 24 April 1996 16:08:47 UTC