Lock Elision / Transaction ABI discussion

[David Collier-Brown]

We hit this in the ARPA days, too.  The possible cases we found were
encoded by Postel as "replies" in RFC 821:

1 - positive preliminary reply, AKA "partial success, continue or abort"
2 - positive completion. AKA "success"
3 - positive intermediate reply, almost the same as "partial success",
but it assumes you must continue
4 - transient negative, AKA "try again later"
5 - permanent failure, AKA "failure"

No-one seems to have used 1, so I implemented it as "information
message" in GCOS SMTP and the arpakit.

I suspect 2-5 are the necessary and sufficient states for
communications, and what we'd now call a pattern...  from which we might
infer a similar, sufficient pattern for locking, one that isn't likely
to paint us into a corner later.

--dave (admittedly feeling somewhat old (;-)) c-b

-- 
David Collier-Brown,         | Always do right. This will gratify
System Programmer and Author | some people and astonish the rest
davecb@spamcop.net           |                      -- Mark Twain
(416) 223-8968