Le 16 oct. 2012 à 08:37, Jan Algermissen a écrit : > On Oct 16, 2012, at 2:09 PM, Anne van Kesteren wrote: >> The algorithm is not for "fixing up". It's for processing URLs, >> including those that happen to be invalid. The end result is not >> always valid per STD 66. > > And there lies the problem. Where is the benefit of producing invalid results > as opposed to fixing with best effort?
I see Anne saying "processing URLs" I see you saying "producing invalid results" > How you implement that is a detail. Not exactly. And it really depends on what kind of issues you have to deal with. On the Web out there, there are plenty of issues in what the browser receive and has to interpret, recover. It's unfortunate, but it is the sad truth. Implementing is then not a detail, specifically for those implementers who suffer from it :) > The important part is that the result is a valid URI. The important part is that the end user behind its browser of choice can use the Web without becoming a QA engineer. For achieving that, we need sometimes to do what we do not want. But in all these discussions, I think it would be more interesting to see actual prose/implementations on how the spec that Anne is writing can be made better more than defining a priori the territories. Basically, let's experiment, and we will see what were good and bad ideas. -- Karl Dubost Montréal, QC, Canada http://www.la-grange.net/karl/
