Getting rid of popup windows for good

Benjamin Smedberg benjamin at
Fri Aug 29 13:24:36 UTC 2014

On 8/29/2014 8:38 AM, Gijs Kruitbosch wrote:
> On 26/08/2014 19:06, Rimas Kudelis wrote:
>> I believe that the ability of a blacklisted website to open pop-ups in
>> response to certain JS events is certainly a bug (I took time to
>> blacklist it, why is that preference not being favored?). If we don't
>> want to block all pop-ups by default, than perhaps at least this bug
>> should be fixed. However, in that case, our permissions UI should also
>> be updated, because this setting is then effectively not binary (allow
>> [all]/block [some] pop-ups), but trinary (allow all/allow some/block all
>> popups).
> At the risk of sounding very ignorant about the product I'm working 
> on... what blacklist is this, and how are you modifying it?
> The popup notification bar, about:permissions, the prefs dialog, and 
> the page info dialog all expose the same system, AFAICT, which is 
> hardcoded to "block most popups by default" (ie based on the events 
> pref you discussed previously), and where changing the pref to "allow" 
> will adjust the whitelist, AIUI.

That's the problem. The current "Block" default only blocks unrequested 
popups, and so if a site has a habit of popping things up on click so 
that they look like a requested popup, there's no way to get Firefox to 
stop that.

In about:permissions terms, we should change the current "Open Pop-up 
Windows" permission from a binary "Block/Allow" to a three-way "Block 
unrequested (default)/Block all/Allow".

> Finally, I expect that there'll be issues where sites both use "good" 
> popups (ie, you specifically click something which tries to open a new 
> window, and if you prevent that from happening, you can't do whatever 
> you were trying to accomplish - things like the dialog you noted was 
> broken, but done by the webpage instead of by chrome, and therefore 
> not "fixable" by us) and bad ones (the ones we *want* to block). 
> Adding a blacklist wouldn't really fix the issue on those sites, as 
> it's an all-or-nothing approach.

I have never encountered a single site which had both behaviors. It's 
certainly not the case that we should be designing for!


More information about the firefox-dev mailing list