- From: Boris Zbarsky <bzbarsky@MIT.EDU>
- Date: Mon, 03 Feb 2003 14:07:10 -0500
- To: Toby A Inkster <tobyink@goddamn.co.uk>
- Cc: "Philip TAYLOR [PC336/H-XP]" <P.Taylor@Rhul.Ac.Uk>, "www-html@w3.org" <www-html@w3.org>
> Possibly we should use a Java packages style naming convention, for > example ActiveX controls could be specified as: > > <object classid=3D"com.microsoft/D27CDB6E-AE6D-11cf-96B8-444553540000" /> > > Whereas plugins for Acme Co Browser would be specified as: > > <object classid=3D"uk.co.acme/1234" /> The issue is that there are actually two existing principal uses of <object> that I am aware of: 1) Using <object> to present content that must be run within a particular plugin binary. This could, eg, be some sort of intranet application or something along those lines. It could be a specific ActiveX control or could be named by some other naming scheme. I am unaware of much widespread use of such <objects>s on the open Web today; natural, they imply rather strict control over what software the people browsing your site use.... 2) Using <object> to present content that can possibly be handled by any of a number of plugin binaries, or even by the browser itself. PDF (which has binary plugins available on at least Linux/MacOS/Windows), Flash (same), Java (not only are there VMs that are available on all platforms, but there are many different VMs available on some platforms), SVG (native rendering in some browsers plus a plugin available) all fall in this category. As XHTML moves forward with deprecating the <img> (and <iframe>?) tag in favor of <object>, category #2 will become ever more important. While a scheme like a uuid that maps to a particular ActiveX object is acceptable for telling the browser how to render <object> files in the first category, the only existing solution to <object> files in the second category is keying off the MIME type of the data to be rendered. I think that the spec (XHTML 2 at least, though an addendum to HTML 4.01 would be nice) should add some clear examples of presenting content category #2 above... In any case, the problem with the naming scheme proposed is the same as what we have with the scheme that exists today -- there is no way to tell exactly what "com.microsoft/D27CDB6E-AE6D-11cf-96B8-444553540000" means unless you hardcode in your browser that "com.microsoft/foo means ActiveX control" (which is what you have to do now anyway to handle the "clsid:" syntax....) Boris -- "Vast quantities of ethyl alcohol (CH3CH2OH) have been discovered near the center of our Galaxy; evidently our Galaxy has also discovered that the best place to store liquor is in the middle of the system!" -- Frank Shu, "The Physical Universe"
Received on Monday, 3 February 2003 14:10:22 UTC