- From: Ian Hickson <ian@hixie.ch>
- Date: Tue, 27 Jul 2010 18:41:52 +0000 (UTC)
On Thu, 18 Mar 2010, Roger H?gensen wrote: > > On my own site currently I mostly replicate the first paragraph of an > article in my journal as the meta description, and write one up for > other pages, usually replicating some of the content. > > I'm both looking for and want a solution to avoid such redundancy. The simplest solution is to just not include a description, and rely on tools to determine automatically what the most relevant information on the page is. > The perfect solution would be a <summary> tag, if you look at the > journal articles on my site you can imagine the first paragraph being > done like this: > > <p><summary>This is just an example, it's a replacement for the old meta > description, and is a brief summary (description) of the page > (content)</summary></p> > > This way the first paragraph in a page would remain unchanged from how > it is done today, and a search engine like Google or screen readers etc. > would use the summary tag instead of the meta description (which is no > longer needed at all in cases like this), if more than one summary tag > the first is considered the page summary one, while the others are > ignored (but still shown as content obviously). That, or an attribute, would be a reasonable solution, but I'm not really convinced the problem is that important. On Thu, 18 Mar 2010, Roger H?gensen wrote: > > Example using HTML5 microdata: (would this be appropriate, would browser > devs, and Google and other search engines support this?) You _could_ use microdata to do this, but I don't think it's really a great use of microdata. This kind of thing would be better done as a microformat, e.g. using a well-known class value. On Fri, 19 Mar 2010, Ashley Sheridan wrote: > > Why not just use server-side code to output the first paragraph of > content as the description for the page also? That is indeed another possible solution to avoid hand-authoring duplicate content. On Fri, 19 Mar 2010, Roger H?gensen wrote: > > http://lists.w3.org/Archives/Public/public-html/2009Aug/0990.html > suggests <link rel="description" href="#desc" />, which is ok I guess. > > But why not simply allow this instead: > <meta name="description" href="#desc" /> > > Existing parsers would notice that content="" is missing which is stated > as being required, parsers that have been updated would notice there is > a href="" instead, so search engines could just look for that id in the > page. I think this would have the highest success rate. > > If backwards compatibility is such a major concern then this could be > done: <meta name="description" content="" href="#desc" /> > > I'm unsure what gives the best result for varous parsers though, would > empty content make them behave the same as if the meta tag was not there > at all? Or would a empty tag cause them to use "" as the actual page > description? > > I'd prefer to have the content attribute missing instead myself, but... <link> is the right element for links, <meta> for text data. Either way, though, the right way to address this is to convince implementors (such as a search engine developer) that they should follow these links and get the description from them. That is an early step in changing the spec: http://wiki.whatwg.org/wiki/FAQ#Is_there_a_process_for_adding_new_features_to_a_specification.3F On Thu, 18 Mar 2010, Roger H?gensen wrote: > > [regarding data-*=""] Maybe a better naming would have been: doc-* It's > short, it kinda reflect what it's related to as well right? Or does that > clash with something? data-*="" is probably too well established to change at this point unless there's a really compelling reason. Cheers, -- Ian Hickson U+1047E )\._.,--....,'``. fL http://ln.hixie.ch/ U+263A /, _.. \ _\ ;`._ ,. Things that are impossible just take longer. `._.-(,_..'--(,_..'`-.;.'
Received on Tuesday, 27 July 2010 11:41:52 UTC