Stef, I'm doing some work in a W3C working group where one of the deliverables is a set of test cases. I.e. a set of machine processable files that give some kind of before-and-after indication of how certain constructs may be processed.
These are used (a) as discussion points for building consensus about exactly what is intended in some circumstance, (b) as a way to go and find out what existing code actually does, and (c) as a secondary document to back up what the primary specification is trying to state. #g -- At 02:11 PM 2/2/02 -0800, Einar Stefferud wrote: >I keep working on Keeping It Simple in honor of Stupid;-)... (KISS) > >In keeping with this, and still seeking some progress, you might note that >my position is reasonably fluid, since the solution(s) do not seem to be >obvious from the beginning. > >It is extremely difficult to do what is needed in the form of Enforcement, >which requires Punishment Consequences and trial courts and all such. All >of which we all agree, should not be mounted or provided by IETF. > >But, let's suppose that someone assembled some documented test cases for >Interoperability, such as were used first for the first pair of >implementations the justified moving a standard from Proposed Standard to >Draft Standard. > >At levels above IP/TCP I suspect that there is very little code required >to do the testing. what is required instead of code is scenarios for >sending this that and the other thing in both directions between >interworking systems. > >I am assuming that such a test was performed at least once, whether >documented or not. I further assume that this could plausibly be used as >an initial Public Standard for testing. This is the specification of the >test, not the code for the test. What kinds of objects are to be >exchanged successfully before that first pair can be accepted as proof of >interworking between that first justifying pair of independent applications. > >I suggest that the first thing to do is stop tossing those test specs in >the trash after they are used, as though they have no further value. >They in fact have the value of a seed that can grow into a valuable long >term testing protocol for all that care about interworking, such that any >customer seeking to buy the most interoperable systems can use the >published test suite protocol to do in-house testing on the systems >offered by bidding vendors. > >So, what I propose is to do something that will give the customers a tool >for protecting themselves from careless or heedless or even dishonest vendors. > >As things are now, we, the end users and customers are basically >defenseless in the face of what appear to be hostile vendors who are >without any checks and balances in the hands of crippled customers. > >If nothing else, our customer community should be interested in founding >an operation that will supply interoperability test scenarios for >themselves. to hell with expecting the vendors to protect the >customers. If the testing tools are not in the hands of the customers, >who can you trust. > >Don't tell me that we should trust the Marketing Droids;-)... >How much testing do those Droids do? > >I suspect they mostly test market savvy, not product reliability. > >But, being suspicious is not a useful thing without some tools to use for >seeking truth. > >I prefer to Trust, but Verify! >This is the power in customer emPOWERment. > >BTW, I do not expect much help from vendors in this strategy. >Though one or two might find some advantage in helping out. > >Especially if they offer real interworking systems;-)... > >Cheers..Stef > > >At 12:22 -0500 30/01/02, Mark Adam wrote: >>Since interoperability on a one-to-many scale would be a problem, >>perhaps approaching it from the many-to-one point of view would be >>better. >> >>Einar's ideas are good, but still difficult to implement. What happens >>when a company fails to find every device it should be tested against? >>It almost seems that what we need is the concept of a reference >>platform. >> >>Having a reference platform allows for a single point of contact for >>everyone wanting "IETF Certification". >> >>I would also suggest that the task of implementing such a platform >>should be up to the WGs creating the standards or the companies >>authoring the standard. This would also give you a group that could >>administer the platform. Of course there would have to be some rules of >>conduct so that nobody could be excluded from performing their >>interoperability testing. (Do I smell a BOF here?) I'm sure groups >>holding reference platforms could find some way to make money off of >>this without breaking the rules. >> >>I'm not saying this would be easy to implement, but it might be worth a >>thought. >> >>mark--------------- >> >>At 00:25 1/29/02, Einar Stefferud wrote: >> >Well now, an idea blinked on here;-)... >> > >> >As Paul Hoffman noted, it costs a small fortune for an entire set of >> >vendor products to be tested against all other interworking products >> >(N**2 pairs is the estimate) and there is no proffered business model >> >for doing this for the entire involved industry.. >> > >> >But, maybe someone can devise a business model for testing one >> >product against all the others that claim to conform to the standard >> >under test. >> > >> >I know that HP did this ounce for their Internet products by hiring a >> >person to do it from one of their customer's sites on the Internet. >> >It does not matter here who or where it was done. >> > >> >But, this puts the burden on the vendors that wish to be able to >> >claim inter-workability with all others, or with some subset of their >> >choice. >> > >> >Or they can identify those that do not interwork for the benefit of >> >those that want to know such stuff. >> > >> >This then becomes an individual company decision, and does not >> >require massed agreement, or require synchronized work schedules. >> >Just put your system on the net and find someone out there to test >> >against. Doing it on the real net is just fine for this testing >> >model. >> > >> >Of course, the vendors that do this can brag or not, as they wish. >> > >> >And here is no great concern for whether every vendor does it or not. >> > >> >And the market can make up its mind by itself. >> > >> >For my view, I have trouble believing that all those vendors are not >> >vitally interested in inter-working among their products. >> > >> >And, in addition, I would hope that someone might mount an open >> >discussion mailing list for people to use to post their private >> >experiences with what does or does not work. >> > >> >And last: This is no longer a useful IETF discussion;-)...\Stef >> > >> > >> >At 09:01 -0800 28/01/02, John W Noerenberg II wrote: >> >>At 10:19 PM -0500 1/26/02, [EMAIL PROTECTED] wrote: >> >>> >> >>>I have in my bedroom a night light, which I purchased at a local >> >>>grocery store. It has a UL logo on it, which doesn't tell me much >> >>>about its suitability as a night light (I can't tell if it's bright >> >>>enough, or if it's too bright, or what its power consumption is), >> >>>but it *does* tell me 2 things: >> >>> >> >>>1) It has been *tested* and found free of any known safety design >> problems. >> >>>It may not *work* as a night light, but it won't shock me when I go to >> >>>throw it in the trash can because it's not suitable. >> >>> >> >>>2) A high enough percentage of night light manufacturers get UL listed >> >>>that I can afford to be suspicious of any company that doesn't have >> >>>the logo on their product. >> >> >> >>Underwriters Laboratories, Inc. is a non-profit corporation that >> >>was founded in 1894. This >> >><http://www.ul.com/about/otm/otmv3n2/labdata.htm>article describes >> >>the process UL uses for developing their standards. Many UL >> >>standards receive ANSI certification. According to the article, UL >> >>relies on information from a number of sources while developing a >> >>standard. >> >> >> >>UL tests products submitted by its customers for *conformance* to >> >>its standards. UL's reputation depends on the rigor and >> >>independence of their testing. I don't know how it costs to submit >> >>a product for testing, but obtaining UL certification isn't free. >> >>UL's certification program is successful, because when consumers >> >>like Valdis (and me) see a UL label, they believe in its value. As >> >>Valdis points out, the value of the label has limits. >> >> >> >>Certification isn't the work of a volunteer organization like the >> >>IETF. It could be the work of an organization like Underwriters >> >>Labs. This would be a good thing for Internet standards, imho. >> >> >> >>One idea proposed multiple times in this meandering discussion is >> >>that those advocating testing should put up or shut up -- create a >> >>testing organization or move on to other topics. I concur with both >> >>those suggestions. I'm sure you'll all be pleased this is my last >> >>word on the topic. >> >> >> >>best, >> >>-- >> >> >> >>john noerenberg >> >>[EMAIL PROTECTED] >> >> ---------------------------------------------------------------------- >> >> While the belief we have found the Answer can separate us >> >> and make us forget our humanity, it is the seeking that continues >> >> to bring us together, the makes and keeps us human. >> >> -- Daniel J. Boorstin, "The Seekers", 1998 >> >> ---------------------------------------------------------------------- >> > > >- >This message was passed through [EMAIL PROTECTED], which >is a sublist of [EMAIL PROTECTED] Not all messages are passed. >Decisions on what to pass are made solely by Raffaele D'Albenzio. ------------ Graham Klyne [EMAIL PROTECTED]