xmodem vs. zmodem

[Dean Gaudet <dg...@arctic.org>]

We all remember the xmodem protocol, right?  Send a packet, wait for
ACK/NAK, send a packet.  It sucked, your total bandwidth was limited by
the latency of the link. 

Then along came zmodem with sliding windows.  Send a bunch of packets,
wait only if the window fills and you haven't received an ACK.  If the
window is large enough to accomodate latency you can get almost the full
bandwidth provided by the link. 

Voting is like xmodem:

    post change
    wait for votes (can take an unbounded amount of time, sometimes takes
		    two or three weeks for obscure patches that most people
		    don't give two whits about)
    if necessary fix problems, and start at the top again
    commit change

Posting and waiting for comments is like xmodem, unless we do something
like "wait for 48 hours and then go ahead unless you've been given a HOLD
ON signal".  :

    post change
    wait for 48 hours (bounded amount of time)
    if necessary fix problems, and start at the top again
    commit change

Committing directly to CVS is similar to that last protocol because it
still allows people to ACK/NAK/discuss the changes. 

    commit change
    if other finds problem with change and has time to fix it, they can
	fix it immediately
    if other finds problem with change and needs you to fix it, you fix
	it asap
    if problem isn't resolved in a reasonable amount of time (i.e. the
	beta cycle if it's a minor problem, or a few days if it's
	major) then revert the change

Note that the latency of this last protocol is the minimum.  It's a lot
more like zmodem.

Draw your own conclusions.

Dean

P.S.  When the whole 1.3 vs. 2.0 discussion came up I said "I veto
anything that keeps us in feature freeze for more than a week".  That veto
wasn't considered valid, so excuse me if I view this entire process as
highly suspect.

Re: xmodem vs. zmodem

[Brian Behlendorf <br...@organic.com>]

At 02:10 PM 7/7/97 -0700, Roy wrote:
>Actually, 2.0 will never get off the ground until a separate module is
>created for 2.0 development, which requires fixing the ownership on CVSROOT,
>which probably requires Marc or Brian fixing the clueless location
>for the cvs passwd file problem first.

Don't let that hold you up... if you need a module created, let me know
what you want done and I'll do it.

	Brian


--=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=--
"Why not?" - TL           brian@organic.com - hyperreal.org - apache.org