Hi, I've been thinking about OBJECT and the various specs/implementations we have, and I've found a few issues, as they cross various groups and are somwhat XAGy aswell, I'm raising them here first - I hope that's ok. The first issues are to do with negotiation: If you wish to use content-negotiatation to serve up an appropriate image format (as the W3 does with its logo) you need to leave the type of object blank, or have it wrong, the disadvantage of having it blank, is that the UA has to make a GET request even to identify the mime-type of the resource somewhat wasteful if you don't want images downloaded for example. I think this one has an easy solution, ensure that content-type can contain a list, and allow wildcards, i.e. in HTML: The other issue with choice in included content is that none seem very powerful, for example in SVG we have the switch element for choosing alternatives, and this is a strict heirachy, use A before B. Often though A and B are of equal quality and it should be up to the user which they would prefer, other times though B is much worse quality and they should only use it if they can't cope with A at all. In http content-negotiation you can do this via the q value - I think it would be very useful to have the same in ML's Consider SVG: Yes Ja Here the two languages are clearly equivalent, however the English will always take priority. However there may be a poor quality translation of the text in second languages and in that situation it would be useful to distinguish it from the first. (I think this applies to all included/negotiated content, not just language choices) I'm not sure of a solution to this one, perhaps simply a q value on each attribute. In HTML, OBJECT is different from embed/image - if I include an SVG in a page with EMBED links from that svg document replace the HTML page, if I do it with OBJECT they replace only the FRAME of the SVG * in fact there's no way to include content which is actually part of the same document as we currently in HTML. To solve this I think we need to have different types of OBJECT, one a way of including content that is part of the Document, and another a way of including a Sub-Document - as they serve different purposes, and are used for different types, consider: somewhere else Here links from the SVG document will open within the SVG "FRAME", the HTML within the HTML "frame", but the jpegs links from the image map will open in the parents window, this difference is somewhat difficult to resolve with different media types. Jim. * Actually it appears they don't where OBJECT works with SVG, however I believe this is wrong as otherwise we have a difference between HTML and SVG which I don't think can be explained, this maybe controversial though.