- From: Dr. Olaf Hoffmann <Dr.O.Hoffmann@gmx.de>
- Date: Tue, 1 Feb 2011 14:13:40 +0100
- To: Ken Stacey <ken@svgmaker.com>, www-svg@w3.org
Ken Stacey: > Olaf, > > Yes, a bug. A long standing bug. ASV, Opera, FF4 and until recently, > Webkit all implement this concept incorrectly. The test suite (author) > interprets it differently. Perhaps the implementations are victims of > the test suite? Until Shane brought it to our attention, it hasn't been > an issue. Well, with some years of experience with my own test suite with about 1400 tests about animation in SVG and the W3C test suites my estimate is about 5% to 10% wrong tests (hopefully only 5% for my own test suite ;o) The advantage of the W3C test suites is, that another member of the working group reviews the test. In my test suite I have to review myself after some time, what is less effective (of course, others are invited to review as well ;o). The problem in both cases is, that the review is not independent from the creation of the test and the reviewer shares experience, interests and interpretation with the author of the test. The advantage of my test suite is, that I can fix known bugs immediately. For the W3C test suites there is some longer delay - and once a test is accepted, it is more difficult to fix to avoid accidental pessimisation of correct but complex tests by other members. And I know that several fixes are simply forgotten over the years before there is a new publication - this is the disadvantage of shared responsibilities and changing interests and members in such working groups. Therefore sure - if one believes in the test result without trying to get to the bottom of the inconsistency with a viewer or the recommendation, there is a high risk to be a victim of unfixed bugs - but typically most bugs result in strange behaviour under slightly different circumstances as in our example here - therefore sooner or later this is discovered finally - and hopefully fixed some day. In this case, my estimate is, that the adobe plugin and the related example in the recommendation are the source of the problem and the test was written to align with this and not with the recommendation definition ;o) To resume - it is important to discuss suspicious tests to get a better, independent review to discover and fix unknown or forgotten bugs. Olaf
Received on Tuesday, 1 February 2011 13:14:18 UTC