Re: Add x86 embed target makefile frag

[jtc@redback.com (J.T. Conklin)]

>>>>> "Andrew" == Andrew Cagney <ac131313@cygnus.com> writes:
Andrew> "J.T. Conklin" wrote:
>> Not until now.  I didn't know anyone was going to implement the Z/z
>> packets until the problems with lost and duplicate packets had been
>> resolved.  Until they are, stateful commands like Z/z can not be
>> considered reliable.

Andrew> I think that the two issues are separate.  Z/z packets can
Andrew> work provided the transport layer is reliable (such as TCP).

True.  But it has to be end-to-end reliable.  TCP is often used to
tunnel to a terminal server, so a TCP connection by itself is not
enough to determine whether or not Z/z are usable.

Andrew> As a separate problem we need to make the remote protocol's
Andrew> transport layer more reliable :-) 

Andrew> If the Z/z packet does prove unreliable then the ``set remote
Andrew> X-packet disable'' command can be used.

I think this needs to be the default.  

Assuming the worst, and requiring people who know it's appropriate
enable it is better than assuming the best and having end users wonder
why their targets aren't behaving (breaking at breakpoints that have
been deleted, etc.).

        --jtc

-- 
J.T. Conklin
RedBack Networks