- From: Chris Haynes <chris@harvington.org.uk>
- Date: Thu, 22 Nov 2001 14:06:41 -0000
- To: "www-html" <www-html@w3.org>
Dear list, Thanks for all the help I've received. I'm not asking this list for help with my coding - that's off topic and totally inappropriate for such an august forum. What I am attempting to do is explain to you the logic leading me to the conclusion that some kind of standardised version negotiation between source and browser is becoming necessary if there is to be a rapid migration to the more recent standards. Some of you suggest I should be able to write 'compatible' code - spanning HTML4.01 and XHTML1.0. Syntactically, this is so (and I had already worked out how to do this with the aid of the excellent XHTML1.0 documentation and validation suite you have produced). The problem actually comes with controlling on-screen layout. I have not found a single way to describe the layout I need (align, valign, margin-size etc.) which is correctly rendered by both pre-and post-generation 6 browsers. HTML4.01 still permits me to use tag attributes like 'align' (which is what I need to use for the 'legacy' browsers as their CSS support is inadequate). XHTML1.0 requires me to use the CSS2 equivalent styles (which I have done - and works fine in the Generation 6 browsers). It is this discontinuity in layout control which requires me to distinguish whether or not I am generating code for an XHTML1.0-capable browser (given that my policy is to use the latest viable standard wherever possible). Since it appears that no one seems close to specifying a workable version-indication standard, my options seem to be: 1) Stick to generating HTML4.01 for the next 5 years or so until all my target market has migrated to generation 6+, 2) Continue parsing the UserAgent strings and maintaining my own 'compatibility' register. (BTW, I parse all the UA info and decode the aliases and morphing undertaken by Opera and MSIE - I don't just look at the 'proper' browser type). I can see that considerable attention and ingenuity is applied to the standards to take account of migration, and it must be frustrating that browser vendors do not seem capable of keeping up. Sadly, in this 'real world' it is the actual browsers, not the standards, that people have to work with. I don't believe that the 'purity' of standards is compromised by making provisions for the inability of the vendors to be accurate, comprehensive and timely, or of the markets to adopt every innovation as soon as it is available If I may observe, as a professional engineer with 35 years industrial experience in the defense, computer and telecomms industries, it appears to me that the lack of support for negotiating version levels could soon lead to even greater difficulties for the (X)HTML community. I can't think of any other sphere of engineering which has survived rapidly-evolving interface standards without suppliers and clients being able to negotiate (or at least validate) interface version compatibility. Looking at the work you are doing on XHTML1.1. and 2.0, with other 'goodies' like XForms coming down the track, the scope for unmanageable syntactic and semantic discontinuities seems enormous (much greater that the 'layout' one I have stumbled over). Masayaso Ishikawa has kindly told me about some of the versioning capabilities that have been considered in the past - but not implemented, as I understand it. Need any help? Chris Haynes
Received on Thursday, 22 November 2001 09:08:31 UTC