Re: Atom ConformanceTests results and feedback
* Sam Ruby [EMAIL PROTECTED] [2006-04-24 03:50]: It would be helpful if every entry had a distinct atom:id. And if the tests were valid: http://feedvalidator.org/check.cgi?url=http%3A%2F%2Fplasmasturm.org%2Fattic%2Fatom-tests%2Fxmlbase.atom Yeah, I should fix those. I’ve also been thinking about the suggestions to use `img` tags to make it easy to scan the results quickly. Likewise I’ve been thinking about Gordon Weakliem’s comment on the wiki: I suggest that the tests be documented with respect to the expected results. For example, TitleConformanceTests is perfect: viewing the feed tells you what the expected result is. LinkConformanceTests, OTOH, gives me no idea of what the author expected to see when viewing the entry in a reader. For example, what does the author expect to see when viewing the second entry? If I display only the second link, do I pass? Do I need to display both links to pass? I’d like to do more, but writing tests is menial work, and I don’t have a lot of tuits at the time being. That’s why I asked about being able to host these at the wiki, so that the touch-up process would be low-friction. If you lack tuits to take care of that, I could copy everything to my site, for the time being, for easier editing. I make no promises as to when any of that will be, though. :-/ Honestly, I’m a little disappointed that not more tests have been written so far, and that is has been happening in such haphazard fashion. Is it really because noone cares? (I suppose I don’t care that much either, judging by my output.) What would it take to get more people more involved? Would it help if there was a list of outstanding testable spec aspects? What aspects need to be tested (this needs more feedback from consumer developers!)? Hmm, #atom would be an ideal place to get this done within a short timeframe, provided a mob got together. Regards, -- Aristotle Pagaltzis // http://plasmasturm.org/
Re: Atom ConformanceTests results and feedback
A. Pagaltzis wrote: Honestly, I’m a little disappointed that not more tests have been written so far, and that is has been happening in such haphazard fashion. Is it really because noone cares? Well the end users probably couldn't care less. The aggregator developers are actively hostile towards such tests. Feed producers probably find informational tests more helpful than conformance tests. And for the poor slobs creating the tests it's just a lot of hard work with no visible benefit. Regards James
Re: Atom ConformanceTests results and feedback
* James Holderness [EMAIL PROTECTED] [2006-04-25 22:15]: The aggregator developers are actively hostile towards such tests. Really? I can only think of counterexamples, though my sample is admittedly tiny. Who are the hostile ones? Personally, as someone who has written patches for an aggregator and is flirting with the idea of building one, I would be very glad to have a defined target to aim at instead of just eyeballing the overlap between the spec and the code. What sort of motivation would compel a developer to be hostile toward tests? Feed producers probably find informational tests more helpful than conformance tests. But they are the ones who stand to gain from consistent and complete implementation of the standard, in the long term. In any case, can’t we even rally four or five people from the WG who care enough about the spec to want to do something likely to increase the chance of good implementations? Where are those who participated in the interminable flamewars brought on by every rathole that lay on the way to RFC4287? Have they stopped caring now, or was all that vitriol just bikeshed painting after all? Regards, -- Aristotle Pagaltzis // http://plasmasturm.org/
Re: Atom ConformanceTests results and feedback
A. Pagaltzis wrote: The aggregator developers are actively hostile towards such tests. Really? I can only think of counterexamples, though my sample is admittedly tiny. Who are the hostile ones? I'm certain not all developers are hostile, but I've witnessed enough of that kind of behaviour to put me off getting involved in this kind of thing. Maybe I'm just being overly sensitive. If you haven't noticed or it doesn't bother you that's great. What sort of motivation would compel a developer to be hostile toward tests? The fact that you're putting a bunch of little X's next to their product. And to make those X's go away they've got to take time off from whatever they're currently working on (which is probably quite important to them) so they can deal with your complaints (which in many cases will involve obscure features of the spec that they'll never encounter in the wild). Feed producers probably find informational tests more helpful than conformance tests. But they are the ones who stand to gain from consistent and complete implementation of the standard, in the long term. That's the thing. Conformance tests will never provide complete coverage of the standard, because large portions of it are completely optional. They need informational tests that can tell them whether the optional bits have any degree of support or whether they're better off avoiding those features. Regards James
Atom ConformanceTests results and feedback
http://planet.intertwingly.net/AtomConformanceTests/ It would be helpful if every entry had a distinct atom:id. And if the tests were valid: http://feedvalidator.org/check.cgi?url=http%3A%2F%2Fplasmasturm.org%2Fattic%2Fatom-tests%2Fxmlbase.atom - Sam Ruby