OT: Arcane question #4324 - "Stateless"

[sargasso]

I have googled and wiki'd my forefinger off.

Does anyone have or can point to a good, unambiguous, correct definition of what "stateless system" actually means?

To be more precise, is there any definition that could be used as the "basis" of a compliance statement?  

I am beginning to think that, for any given transactional system, the term is actually meaningless.  At least above the level of transmission protocol it is.  The "best" I have come across so far is at http://whatis.techtarget.com/definition/0,,sid9_gci213051,00.html
However, IMO, even this is not technically correct - at the level of http state information is held at both nodes via the session concept.  

Why do I ask?  It is a "requirement" of the current assignment that the system be stateless.  And there are problems arising due to an asynchronous handoff to a third system.  If the term is essentially meaningless then I can forget the semantic excercise and get on with "fixing it".

tia
bruce

[thomaskilian]

Hi Bruce,
why don't you ask the guys requesting that requirement to put it down in a glossary? A stateless transaction system is a contradictio-in-adjectio (like a tame beast).

[sargasso]

Thanks TK, that's what I thought.

It's my own dumb fault for taking on a ($EXPLETIVE) compliance audit assignment I suppose,  at least its only a quicky.

bruce

p.s. I shall refrain from asking them, the last thing I want is more ambigous drivel.  

[«Midnight»]

...and get on with "fixing it"

Bruce,

Which meaning of the term "fix" did you have in mind?

• To make permanent and unchanging?
  
• To corrupt?
  
• To neuter?

There are other meanings, but they don't seem to have any relevance to software...

[sargasso]

A:

To neuter

My observation that the system did not appear to be "stateless"  as the developers are :
a) creating a logical session (above and beyond the ssh session already in place), and
b) holding transactional information within that logical session, including inventory locks,
raised a "Yes it is, no it isn't" level argument between the develpers and the "architects".

The system semi-adequately cancels the session and releases the inventory after a timeout (which is again IMO stateful behaviour).  But, because of the asynch 3rd party hand off, there is a possibility that the session could be cancelled in an invalid manner (customer is charged but no goods delivered) or never cancelled (supplier-3rd party interaction keeps going long after the customer has abandoned the transaction) causing inventory to remain locked.

I am trying to stop the said kindergarten argument and have one or both of the "children" accept that there is a problem and take appropriate action.

Having noted that desexed pets tend to be less territorial and more placid, I guess that's the best course of action here!!!

regards
bruce


p.s.  As this is OT, and its only a very short term issue for me (final report Tue) we can end this thread now.