- From: Tantek Çelik <tantek@cs.stanford.edu>
- Date: Mon, 12 Apr 2004 10:37:55 -0700
- To: "Anne van Kesteren (fora)" <fora@annevankesteren.nl>, Ian Hickson <ian@hixie.ch>, Boris Zbarsky <bzbarsky@MIT.EDU>
- Cc: <www-style@w3.org>
On 4/12/04 9:41 AM, "Anne van Kesteren (fora)" <fora@annevankesteren.nl> wrote: > >>> Compare: > >>> h1 { content: url(xyzcompany-logo); } >>> h1 { content: "Logo: " url(xyzcompany-logo); } >>> h1 { content: "\160 " url(xyzcompany-logo); } >>> h1 { content: "\feff " url(xyzcompany-logo); } >>> h1 { content: url(something) url(xyzcompany-logo); } > >> Do you have a better proposal? I haven't fully thought this through yet. > > I think that when image within a "serie" could not be found, the next > "serie" (separated by commas) should be used. If none such serie exists, > as with all the examples from Boris, "contents" should be used instead > (as final fallback). Yes, such an "all or nothing" approach to the value makes sense to me, and I think is in the spirit of how CSS values are supposed to be handled, that is, if a UA can understand and handle the *entire* value, it should use it, otherwise it should ignore it. A URL that fails to load (or is of a type not handled by the UA, or is too big for the device etc.) seems to fall into that category. Perhaps this should be a consideration for all CSS properties that have some form of URL among their valid values? E.g. CSS1: - background(-image) does not have "all or nothing", no explicit fallback (the color is *always* rendered under the image, thus it isn't a fallback per se). It might be interesting to be able to say: background: url(bigimage.tiff) silver no-repeat 0 0, url(smalltileimage.png) white, gray; - list-style(-image) does not have "all or nothing", explicit fallback to list-style-type CSS2.1: - content does not have "all or nothing", no explicit fallback - cursor *does* have all or nothing, explicit fallback via comma delimited list CSS3 UI: - cursor *does* have all or nothing, explicit fallback via comma delimited list - icon *does* have all or nothing, explicit fallback via comma delimited list (will be in CR draft). etc. Tantek
Received on Monday, 12 April 2004 13:42:44 UTC