- From: Charles McCathieNevile <charles@sidar.org>
- Date: Sat, 19 Nov 2005 16:45:21 +0100
- To: "Jim Tobias" <tobias@inclusive.com>, "'Patrick Lauke'" <P.H.Lauke@salford.ac.uk>, "'WebAIM Discussion List'" <webaim-forum@list.webaim.org>, w3c-wai-ig@w3.org
On Fri, 18 Nov 2005 14:53:18 +0100, Jim Tobias <tobias@inclusive.com> wrote: > Is there a "best tool overall", or rather a (small) set of tools, each of > which is best at one part of automated testing? I'm assuming the latter. I think there are different tools that are good at different testing tasks. Amongst them, I like Hera's guidance for manual testing, its EARL reporting, its simple support of collaborative testing, its translation interface that makes it easy to adapt to your language, its reporting style, and its philosophy. Although as one of the people involved in its development, that is hardly surprising. I like AccVerify for its raw power - in dealing with a large-scale site professionally I wouldn't want to be without it - the full on paid version, for professional work. I like the Wave for its quick overview - as an 'expert' trying to get a handle on what a site is like, the Wave is often the first tool I reach for. (Hera or AccVerify are next). > Their other criteria are reliability and ease of use. Their goal is to > split web accessibility into two parts: automated testing performed by > relatively untrained testing technicians, and complex testing performed > by highly trained usability/accessibility engineers. I can't fault > their goal; can you? Nope. The idea of making a McDonalds industry out of basic testing, bubt actually using real expertise for the stuff that is difficult makes good economic sense. > Given WAI's defensible "no recommendations" policy, there seems to be no > coordinated public source for good general guidance on automated tools. Well, any recommendations from WAI would in fact be indefensible. However, if they had the resources avalable (they are looking for a technical person just to maintain working groups, and I encourage anyone who wants to work there to apply, since I am suffering from working groups not having any support) they could do more testing, and publishing of results. > At the least, there *should* be an agreed-upon list of which guidelines > can be robo-tested, and which tools perform satisfactorily. I can't > find such a > resource; am I missing something? I worked on this in EuroAccessibility - unfortnately it seemed that many of the other members were more interested in securing funding for themselves than in doing any real work, so the Task Force was closed. But have a look at http://www.euroaccessibility.org/tf2_doc/method/evalmeth.html - the draft we got to, with rreferences to Giorgio's work that was an important basis. It turns out to be very difficult to get agreement on which tests can be reliably performed automatically - there are disagreements on the finer points of testing, not everyone is prepared to release their testing algorithms (Hera and aChecker are open source, so you can look at the tests they apply), and making claims about the technology people can or cannot develop is in my experience a fairly difficult thing to do. Giorgio Brajnik (you should learn his name if you don't know it - he is involved in LIFT, a repair tool that in my opinion shows up all the others for usability, although it isn't my first choice for its testing algorithms) has done some interesting work on methodologies for comparing tool results. If there is interest in putting some resources (effort) into benchmarking tools, I will try to resurrect the EuroAccessibility Task Force in the new year or make something similar. It is certainly true that this information would be helpful. It seems that the people running off with the funding grants are generally not doing anything much about providing it :-( cheers Chaals -- Charles McCathieNevile Fundacion Sidar charles@sidar.org +61 409 134 136 http://www.sidar.org
Received on Saturday, 19 November 2005 15:45:58 UTC