A while ago I suggested a different packaging system for add-ons that would (arguably) make it easier to work with. I proposed it as a project for the Google Summer of Code, and it was accepted. That was kind of unexpected, and I was very excited to be part of that experiment. There was a good student submission for the project and everything looked well underway.

Fast forward a few months, and I was contacted by the student working on the project. He wanted to drop it due to lack of time and progress . After chatting for a while, I agreed with him that dropping it was for the best. In the end the problem was a mixture of poor communication and a scheduling mismatch between the summer break in the US and other countries (Brazil in this case). He just couldn’t focus on the project due to school work.

So, the experimental implementation never happened, but the idea lives on. The project’s approval motivated me to create a pretty decent spec that I’ll update once Firefox 4 is released. Many changes have been introduced in Firefox 4 that affect manifest files, and others that overlap with some of the proposed features. Once Firefox 4 is stable for add-on developers, I’ll update the wiki and probably post an update here. I don’t have any plans to implement this or push the idea too hard, but I still think there’s some value there and it’s worthwhile to keep it alive.

On Thursday, the update queue had about 80 add-on updates in line to be reviewed. This is a short as it’s been for a very long time. Today it stands close to 200 updates, and will continue to grow in the following couple of weeks. All authors are now rushing out updates because they just realized Firefox 3.6 is out.

Well, now it’s up to the editor team to catch up. It’ll take a while, I think. Authors that decided to update after the 3.6 release will now have to wait. It’s bad for everyone, and the result of bad communication. That’s something we’ll need to work on.

First of all, let me be very clear about this: there is no short term plan to eliminate the extension platform as we know it. XUL and XPCOM run deep in Firefox. They are Firefox. Eliminating the technologies that make extensions possible would require a rewrite of pretty much everything in the platform, which is massive. I’m not saying this couldn’t happen some time in the future (I don’t know, really), but it isn’t something that can be accomplished within a few weeks, or even a few months. It’s something that requires a great deal of planning and the collaboration of the whole community. Extension developers shouldn’t worry about their add-ons being obsolete overnight.

We should all look into the future, though. Not as something that we’ll have to accept, but as something we can shape. Jetpack and Personas are still experiments in many ways, and there’s much we can do to make them what we want them to be. I personally doubt they will ever reach the point where they will replace the current add-on options, but I am confident that they can come very close, and that’s a big win for everyone.

Jetpack

The goals of the Jetpack project are ambitious: no restarts for install / uninstall, a clean and more stable API, complete security, and a much easier development experience. They’re so ambitious that to think all of these can be accomplished while preserving the flexibility of the current platform would be naive at best.

There are no stable APIs. You can make higher abstractions that are less likely to change. But they will change. Jetpack only makes its add-ons dependent on its API, instead of the XUL/XPCOM platform. So, instead of updating your add-on to the next Firefox version, you’ll update it to the next version of Jetpack, which should happen much less often. That is of course assuming there will be some sort of versioning of the Jetpack API. If that’s not the case, well, then we have bigger problems to be concerned about.

Jetpack, unlike XUL and XPCOM, is not a fully open system. It can’t be. Not without sacrificing the security it’s meant to bring. As a secure system, it should be closed by default, enabling through its API only the features that are considered to be safe and necessary. This limits add-on creativity to the API designers’ imaginations, as opposed to the developers’. On the other hand, in the current system extensions can do pretty much anything. They can read, write and execute files. They can change your preferences and access your saved passwords. They can monitor your online activities and send information to third parties. The only real protection between you and the extensions you install is the review system that all of them have to go through in order to be publicly listed on AMO. A group of reviewers (also known as editors) make sure these extensions are safe to use and respect user choice. Which one is better? To me, the answer is simple: if it’s possible on Jetpack, use Jetpack. It’s simpler and safer. If it isn’t, then fall back to extensions, where you have almost limitless control.

Jetpack add-ons will be easier to develop, and it will be possible to install and remove them without restarts. This is a huge win for users and developers. Many, if not most add-ons will be easily portable to this new platform, and they will benefit from it. It remains to be seen, however, if highly complex (and extremely popular) add-ons like AdBlock Plus, NoScript and Firebug will be able to live in the Jetpack world. These extensions are strongly tied to the platform, and their interactions would be very hard to translate into a general use API. Maybe we can implement jetpack.magic.doWhatNoScriptDoes .

Personas

Personas are not even close to being a replacement for themes. Personas allow some very basic skinning using header and footer images and setting font colors for the main toolbox. Surely they could be extended to include images for the toolbar icons and some more advanced customizations, but that’s not the case now, and even then they wouldn’t be a complete replacement for themes. Themes can change the appearance of the application in very significant ways, and this can’t be accomplished without the complexity inherent in theme development. So, theme developers, you’re not done yet either.

The future

I think the future for extensions and themes is still bright. Jetpack and Personas have shown us how things can be different, and opening the field for new development and competition is a win for all.

There’s much we can do to improve the “classic” add-on world, in the area of documentation and tutorials, and even in the platform itself. We’re being shown how to improve. We should take this as a call to action, and improve. Let’s work on the platforms that will support the development of the future, but let’s not forget the ones that are still active and thriving.

Let’s not forget how we got here.

This discussion came up due to a denied update for Text Formating Toolbar 0.1.4.11. I was requested to come up with a POC that demonstrates how name conflicts can cause problems, so I created this add-on, which conflicts with Text Formatting Toolbar using seemingly harmless code. Here’s the link:

Name Conflict

If you install this with the Text Formatting Toolbar, you’ll notice that the bold, italics, code and color features are broken.

Here’s the code in my overlay:

window.addEventListener(
  "load",
  function() {
    window.setTimeout(
      function () {
        formatItalics = "i";
        formatBold = "b";
        formatColor = "white";

        formatCode =
          function(aString) {
            if (typeof(aString) != "string") alert('Invalid Code!'); }
      },
      100);
  },
  true);

As you can see, it’s fairly simple. The load event and setTimeout call are only used to make sure that it is the other extension and not this one the one that experiences the compatibility problems. If two add-ons use the same names in the global scope, then one is going to experience problems in one way or another. There’s nothing more to it.

Obviously I can intentionally create an add-on that causes conflicts with any other add-ons, but I hope you see past this and realize that the code I posted could very well be written by another author who is unaware of the rules we enforce. This is what we look after, and this is why we should reject global namespace pollution in almost all instances.