- From: Dr. Olaf Hoffmann <Dr.O.Hoffmann@gmx.de>
- Date: Fri, 26 Sep 2014 12:10:32 +0200
- To: www-svg@w3.org
Hello, if the HTML5 draft (or another draft not yet applicable for SVG) has restrictions about what happens with content of an img element and this does not apply for object elements, authors needing such external file referencing should simply use object, not img - the concept of img was already somehow outdated in HTML4 compared to that of object. Questionable why the the HTML5 draft introduced a new element embed with a similar poor concept, but no use case at all. In some areas of HTML5 really look like a continuation of the unfortunate HTML 3.2, sticking in the last century - not something useful for today. And if object has a method to care about privacy, why img does not? What about the drafts about HttpRequest - intended to gather and sending information due to scripting without a need to inform the user about this? If we combine this with WebStorage, this creates huge holes in the concept of privacy - why to specify/implement such things, but to worry about privacy for the img element? Mozilla worries about declarative interactivity without scripting in SVG dure to security holes in HTML5 form features, but does not worry, if script interpretation is activated - obviously the idea is, that people with activated script interpretation do not care about privacy anyway, but people with script interpretation switched off may care? All this sounds like a crude concept of privacy in current viewers/efforts of the W3C - would be better to rethink this concept, if there is one at all, than to obfuscate existing recommendations ;o) If HTML5 does not manage such issues of this format, it is obviously not ready to become a recommendation... Obviously, both the (X)HTML file and the embedded SVG file can reference the same (alternate) stylesheets. The main problem occurs here by switching between styling alternatives. Most viewers will not switch automatically to an alternative style with the same name in a referenced file, some do not at all allow to switch to alternative styles - for me this is the core problem here, but these are more implementation gaps, no specification problems in SVG. To apply styling from the embedding file to a standalone embedded file seems to be a very bad idea, this would be a feature following the 'principle of most surprise' for skillful/informed authors - and might result in a lot of nonsense presentation, especially if we take into account the idea for SVG 2 to convert lot of attributes to properties together with the typical viewer problem, not to care about format version indications in documents. Soon this can turn in a feature of 'most surprise' for the whole audience. Olaf
Received on Friday, 26 September 2014 10:11:02 UTC