Le 25/08/2010 01:15, Ian Hickson a écrit : > On Wed, 25 Aug 2010, David Bruant wrote: >> Note after "10.1.1 The DOCTYPE" : >> "DOCTYPEs are required for legacy reasons. When omitted, browsers tend >> to use a different rendering mode that is incompatible with some >> specifications. Including the DOCTYPE in a document ensures that the >> browser makes a best-effort attempt at following the relevant >> specifications." >> >> Reading the following description, I have the impression that HTML6 will >> require a DOCTYPE as well. >> >> Following the hypothesis that the HTML language or HTML parser >> definition may change, it means that HTML6 may need a different DOCTYPE >> than HTML5 in order to "[follow] the relevant specifications". > It's likely that there won't be an HTML6 -- "unlikely" (likely-won't) doesn't mean that there won't be. The eventuality should probably not be thrown away that easily. Twenty years ago, who could have predicted what happened until today ? Can we already predict what is going to (not) happen twenty or one hundred years from now ? I think that this statement should be very carefully justified.
> not because HTML5 will be the > last version, but because we've moved to an unversioned development model > where there is simply a continually-maintained HTML specification that is > always current. If, when you say "we", you mean "the WHATWG and the W3C", this is incoherent with what is written here : http://wiki.whatwg.org/wiki/FAQ#When_will_HTML5_be_finished.3F "It is estimated by the editor that HTML5 will reach the W3C Candidate Recommendation stage during 2012." What is incoherent is the idea of reaching the "W3C Candidate Recommendation stage" or even reaching a stage for the overall specification. If "we" means "WHATWG", then ignore my previous point. However, would the W3C and WHATWG models be compatible in that case? Can both work on the same spec and at the same time for one having a mature enough specification to say "this is a Recommendation" and the other saying "this is still in progress" ? If the continually-maintained model is the one currently chosen, does it really make sense to call the specification HTML5 ? Shouldn't it be HTML ? I agree that the naming question is not very central. However, changing the name would be a good way to advertise the idea that the work on the spec won't end. I think that a lot of people are expecting HTML5 to be eventually a finished specification as HTML4 was, as actually all specifications I can think of have been. How can a HTML validator be written ? Should it be continually-maintained as well, making correct document incorrect and vice-versa depending on the validation day ? I guess that the changes between HTML4.01 and HTML5 will be continually-maintained as well ? >> What will this doctype be since it cannot be <!DOCTYPE HTML>? > It can be that. HTML is backwards-compatible, meaning that an > implementation of the spec in 2020 will handle content written to the spec > in 2010 correctly. Even if I agree on this goal, I think that this is a very ambitious statement. >From a formal point of view, how can you prove that a change that you make on a spec is backward-compatible with *any* content written following the 2010 spec? If I was asked to guess what kind of problem this is, I would say "undecidable". I'm an humble computer science student, though :-) In my opinion, if changes are imposed to be made with retro-compatibility in mind, they are going to be always more and more difficult and take more and more time to be proven (formally or informally) retro-compatible. Versionning seems to have this advantage of having a back-door to say "the previous version was good but had this issue. The new version is like this which corrects this issue". I have another concern about educational material. If I want to write materials on HTML5 now. You seem to guarantee that if I do it right, the content that people would write based on thess materials will be correctly interpreted which is good. However, no one can tell how up-to-date the informations on that material will be a century later since there is no real trace of the differences during one point in time and another. Or is there ? David