[whatwg] borders on images inside links

[Ian Hickson]

On Thu, 8 Apr 2010, Ashley Sheridan wrote:
> On Wed, 2010-04-07 at 17:45 -0700, Tab Atkins Jr. wrote:
> > On Wed, Apr 7, 2010 at 5:08 PM, Steve Dennis <admin at subcide.com> wrote:
> > > On Thu, 8 Apr 2010, Ian Hickson wrote:
> > >>
> > >> Personally, my opinion is that images in links should have borders 
> > >> because otherwise how do you know it's a link?
> > >>
> > >> This seems to be a minority view, though. People have been 
> > >> explicitly turning off this cue for literally over a decade.
> > >
> > > I think that browser defaults should make sense when the page is 
> > > rendered Without author styles.  While authors often override this 
> > > particular feature, there they have the option to represent linked 
> > > images in other ways such as hover states, or more attractive 
> > > borders etc.
> > >
> > > Authors also often overwrite many other browser defaults such as 
> > > font, styles for horizontal rules, often margins on 
> > > paragraphs/headings/lists etc. But it's probably not a good idea to 
> > > set these things to zero.
> > >
> > > At the end of the day, I've never found turning off borders much of 
> > > a hassle as an author.
> > 
> > As Ian said, though, borders are turned off *so often* that I don't 
> > think it's an actual cue to users.  The actual cue I, and many normal 
> > people I know of, use to see if an image is a link is to put the mouse 
> > cursor over it and see if it turns into a pointer.  That's still 
> > present, so we should be good.  It's not ideal for mobile browsers 
> > without an explicit pointer, but I get along fine on my phone.
> 
> A point that Ian also noted was where styles were turned off, or the 
> page was viewed in a browser not capable of displaying styles. Yes 
> they're rare, but they do exist. As he said, it's not a huge effort to 
> turn them off in a stylesheet, and is often just one of many things 
> people 'reset' along with table cell padding, header margins, font 
> sizes, etc. These are among the most often styles that I see people 
> typically change, and it's usually the same kinds of changes. Should the 
> browsers all change their behaviour to follow what is the trend of 
> styling now, or remain historically consistent and let the developers do 
> a little more legwork.

The thing is, people have been turning it off with "border=0" for a long 
time, to the point where even non-CSS browsers don't reliably show this 
cue. I don't think it makes sense to insist on this (superior, IMHO) 
rendering, in the face of such frequent author disagreement.


> There is another side to this as well: what bandwidth impact might this 
> have if stylesheets online didn't have to consistently change the 
> default behaviour for popular elements?

The impact here would be minimal either way.

-- 
Ian Hickson               U+1047E                )\._.,--....,'``.    fL
http://ln.hixie.ch/       U+263A                /,   _.. \   _\  ;`._ ,.
Things that are impossible just take longer.   `._.-(,_..'--(,_..'`-.;.'