- From: Leif Halvard Silli <lhs@malform.no>
- Date: Fri, 19 Sep 2008 00:19:37 +0200
- To: Lachlan Hunt <lachlan.hunt@lachy.id.au>
- CC: hallvord@opera.com, public-html@w3.org
Lachlan Hunt 2008-09-18 22.33: > hallvord@opera.com wrote: >> I haven't checked the spec but this implementation for _blank isn't >> going to be compatible with the web. I know because we (Opera) tried >> to not set window.opener for a new window created by a _blank link - >> it broke Google Docs' document manager screen interaction with the >> opened documents. > > I tested this in Firefox, IE and Safari, and they also set > window.opener, so this behaviour will need to be specced. I did check > the spec, but I find anywhere that covered this issue appropriately. The way the draft currently is written, the "_blank" is covered in the 5th list item of Section 4.1.6 which says: "Otherwise, a new browsing context is being requested, and what happens depends on the user agent's configuration and/or abilities: [etc]" [1] I have not tested Google Docs for this yet, but the problems with my Internet Banking that I mentioned was in the current iCab. Is it possible for users to set up [without any third-party plug-ins] the "big four" Web browsers in a way that makes them incompatible with e.g. Google Docs? >> If this is in the spec we had better drop it - perhaps a brand new >> value like <a target="_standalone" href="..."> could do the job but >> we can't change _blank. It seems to me that such a "_standalone" could be defined as a "_yes-I-really-mean-_blank". > We don't need a new value for this because no-one has demonstrated that > there is a real problem here that needs solving. Until that happens, > coming up with potential solutions really isn't worthwhile. It sounds from what you and Hallvord said above that you think _blank really is (and should be) what Justin called "_fork" or "_continuation". [2] Another place where this is relevant is @longdesc. I believe all UAs with enabled support for @longdesc will open the @longdesc link in a new Window/browser context above the current one - and not in a Tab. And when you close that window, you're back to the originating window. This is an important effect of using @longdesc. Note that in the @longdesc case, it doesn't matter - as much as I can see - whether there is a formal "link" between the windows. It is enough that the @longdesc URI opened in a new Window. Close it, and you're back. You and Hallvord are mostly occupied with the "link" between the originating context and the context created by the _blank link. But it is also important to have an eye for the usability aspect. It is not enough that my browser, which opened the calendar picker in a Tab instad of a Window, only inserts the correct date into the originating page. It should also bring me, the user, back to the originating page, with 100% certainity. Or else, it should not open a Tab at all, but instead open a Window. And that is whey I care about whether the page actuall opens in a Tab or in a Window. What is the relationship beween browsing context and Tab? [1] http://www.w3.org/TR/html5/web-browsers.html#the-rules [2] http://lists.w3.org/Archives/Public/public-html/2008Sep/0376.html -- leif halvard silli
Received on Thursday, 18 September 2008 22:20:22 UTC