Living in a Go Faster, post-XUL world

Eric Shepherd (Sheppy) eshepherd at
Fri Jul 17 02:44:27 UTC 2015

Absolutely. There are lots of things that can be parallelized here. Some of the work is obvious, as you point out: redoing basic presentations from XUL to HTML. Figuring out where HTML5 isn't up to the task can be done alongside that work, and we can propose new standards to add the stuff we need to take those next steps.


From: Gijs Kruitbosch <gijskruitbosch at>
Sent: Thursday, July 16, 2015 5:18 PM
Subject: Re: Living in a Go Faster, post-XUL world
To: Eric Shepherd <eshepherd at>
Cc: Firefox Dev <firefox-dev at>

           Converting our in-content pages, converting our web-facing XBL (video controls come to mind), investigating more or less orthogonal perf improvements for startup (potentially by replacing some bits of XUL with HTML in the main window, but maybe through other means), improving "new" CSS flexbox performance, making an inventory of "XUL things" that we rely on right now that don't have HTML equivalents (XUL persistence, window attributes that get reflected to native code, certain -moz-appearance values, layout for panels/popups where the bounds are determined by the page instead of the bounds being determined for the page, maybe other stuff...) and figuring out paths to replace those, expanding exposed APIs (like e.g. CustomizableUI) to accommodate other things that add-ons currently do with XUL overlays (e.g. "load script X in every browser window", "add menu Y in the Tools menu", "add fancy toolbaritem (not a button) <whatever>" to the set of customizable toolbar items... etc.    
          All of these can be worked on more or less in parallel. There's probably other things that I've forgotten about.    
   ~ Gijs   
       On 16 July 2015 at 22:07, Eric Shepherd     <eshepherd at> wrote:    
                David Rajchenbach-Teller wrote:
                 P.S.: I can clarify "finite" as "finished long, long before we switch toServo", if you prefer.         That helps... a little. :)      
 I think it's clear (I *hope* it is, anyway) that we need to find ways to do this iteratively, instead of it having to be a single huge, monolithic project that will essentially never be finished.      
 Figuring out what those iterations are is the trick.      
         Eric Shepherd        
         Senior Technical Writer        
         Check my Availability        
 firefox-dev mailing list     
      firefox-dev at  
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <>

More information about the firefox-dev mailing list