Add-on-Con Wrap-up

Yesterday, despite the best efforts of a snowstorm in the South, I made it to the first ever Add-on-Con hosted at the Computer History Museum about a minute or two away from Mozilla’s office. The event exceeded my expectations and was a great experience.

It was a weird, great feeling being around so many people that spend their time working on add-ons and care about them as much as I do. Some of my thoughts about the sessions:

• In the opening keynote panel on add-on business models, it was great to hear that our compatibility outreach with Firefox is helping developers and that being featured on AMO has such a positive impact on extensions.
• Brian King gave a great talk on the state of the Mozilla Add-ons universe, covering the history of extensions in Mozilla, AMO, Mozdev, and a number of other topics.
• I was unexpectedly part of the Add-on Distribution Strategies session panel, along with Rey Bango (Mozilla), Pat Buckley (WebMynd), and Alec Jeong (Cooliris). It was interesting to hear and talk about the journeys add-ons go through for better distribution and exposure, and reminded me how important metrics and statistics are to making those decisions. I have some follow-up notes on this and will be blogging about this more in the future.
• The closing keynote was a browser vendor panel of Microsoft, Mozilla, and Google. I was very glad that the questions were all add-on related rather than a generic browser panel seen at other conferences.

One of the things that struck me about yesterday was that “AMO” was mentioned in every talk I went to without ever being defined. I usually always define it the first time I use it, but everyone at the conference seemed to know what AMO is, which was a very strange feeling.

Overall, the conference was a really great thing for both business folks and technical folks, and people on the add-on development side as well as people on the browser vendor / distribution channel side, especially for being its first year. I can’t wait until next year.

Impact Mozilla, Thunderbird 3.0b1, and Firefox 3.1b2!

Voting has begun for the Impact Mozilla contest, and after taking a look at all of the finalists, I was very excited to see that about half of them involve using add-ons to promote Firefox and increase retention. Check out the finalists and vote before next Wednesday!

Today, Thunderbird 3 beta 1 was released. I tried it out and immediately switched to using it as my default client. The upgrade from 2.0 was seamless, and I’m really enjoying the new features and interface. Of course, one of my favorites is the new Add-ons Manager, first introduced in Firefox 3. It will now be possible to install extensions from AMO directly in Thunderbird by searching for them in the Add-ons Manager.

Speaking of beta releases, Firefox 3.1 beta 2 was released yesterday with Private Browsing mode, TraceMonkey turned on, and one of my favorite less-announced features: being able to drag the window around on Mac from any part of the chrome (also available in Thunderbird 3.0b1!).

Lots of exciting things so far this week, and it’s only Tuesday!

Add-on-tastic

I’ll be in Mountain View next week to attend Add-on-Con, a conference all about add-ons with Mozilla, Microsoft, Google, and of course add-on developers. It will be a great chance to talk with developers firsthand and take in some interesting sessions.

From Mozilla, Mark Finkle will be giving a presentation on leveraging third party APIs to create mashups with add-ons. Having Mark in attendance is great for both myself and add-on developers: I get to send all technical questions to Mark, and add-on developers don’t get incorrect technical answers from me! It’s a win-win.

Mike Shaver will be taking part in a panel on the future of the web browser, along with representatives from Microsoft and Google to wrap up the sessions. As former Director of Ecosystem Development (loosely translated: “add-ons”), Mike has a long history of working with add-on developers and improving extension support in Firefox. Plus, he’s kinda funny.

On the business track, Brian King will be exploring the state of the Mozilla Add-ons universe. You can read more details in Rey Bango’s post on the Add-ons blog.

Additionally, Mozilla will be hosting an add-ons open house the night before, so if you’re interested, you can find the details on the wiki. Unfortunately, I won’t be able to attend the open house as my flight gets in later.

I’m also very happy to announce that I’ve accepted a full-time position with Mozilla after graduation in a few weeks, and will be moving to the San Francisco Bay Area very soon. It’s been a great 2-3 years being immersed in add-ons at Mozilla and I’m very excited for what’s coming in 2009.

1 Billion Add-on Downloads

This morning, addons.mozilla.org reached a huge milestone: we served our 1 billionth add-on download since we started keeping track of downloads in 2005.

This number only includes downloads from addons.mozilla.org, but as many add-ons are hosted elsewhere, we can be sure the actual number is even higher. This is a very exciting week for Firefox add-ons, with the launch of Fashion Your Firefox yesterday, offering a new way for first-time add-on users to customize their browser.

This amazing feat was made possible by the huge community of add-on users, developers, and enthusiasts, as well as non-Firefox applications that embrace Mozilla add-ons like Thunderbird, SeaMonkey, Sunbird, Komodo, Flock, and Songbird, to name a few.

At 1.5 million add-on downloads per day and increasing, I’m confident our next billion will arrive even more quickly than the 3.5 years it took for the first. If you’re not using an add-on yet, try one out. They’re pretty popular.

New Compatibility Dashboard Live

Just a quick update to yesterday’s post about Add-on Compatibility Tracking: the first part of the compatibility tracker is now live on AMO, so give it a try if you’d like to see detailed compatibility information per add-on. It’s updated hourly, and we’re up to 21% compatibility, +1% in the last day.

The rest of the tool should be ready next week.

Firefox 3.1 Add-on Compatibility Tracking

I recently posted about the start of the campaign to get add-ons ready for the upcoming release of Firefox 3.1, and wanted to explain what our goals are and how we’ll be tracking progress.

As with previous Firefox releases, we’re aiming to have 90% of add-ons that make up the top 95% of add-on usage compatible with Firefox 3.1. That’s not an easy goal to comprehend, so I’ll explain what it means and why we do it.

Every day, the Mozilla Add-ons website gets around 135 million update checks from add-ons installed in Firefox and Thunderbird applications all over the world to determine if any updates are available. While there are some extensions that account for several million of those pings alone, there are many extensions that are not as popular and may have less than 100 active daily users. To make sure our outreach efforts benefit the most end users, we rank the add-ons by their number of active daily users and focus on helping the add-ons that make up 95% of the total add-on usage.

As shown in the above graph, only 861 add-ons as of last Wednesday make up 95% of the total update pings, while 5% is accounted for by the much larger number of 4943 add-ons. Our goal is to get 90% of these 861 add-ons compatible with Firefox 3.1.

For the Firefox 3 effort (Operation Threedom), Alex Polvi created a status bar page that helped visualize our current progress. For 3.1, we’re taking it a step further and integrating a Compatibility Center within AMO that can always be checked for the latest progress.

The new tool isn’t available yet, but should be within a week or so. Here’s a screenshot of our current progress:

As shown, we currently have 20% compatibility. The report is customizable by version, so the current Firefox 3 version of the chart shows 94% compatibility. More screenshots and progress can be found in bug 460309.

We’ll be posting periodic updates on the progress of the Firefox 3.1 compatibility push to the new Add-ons Blog and in various other add-on communication channels.

Responsible First-run Usage

Extension first-run pages are increasingly common and are a great way to inform the user about what the extension they just installed can do, how to access it, how to get help, and oftentimes guide the user through creating an account or logging in to a service. They are usually very graphical and clean-looking, and really show off the extension or author’s brand. Sometimes in addition to or in lieu of a first-run website in a new tab, additional first-run components are used, such as opening a sidebar or popping up a modal dialog or wizard.

The practice of using first-run modal dialogs, pages, sidebars, and wizards is a great way to make sure the user doesn’t forget to try out your extension after installing it. But there’s a pretty big problem that’s developing as more and more extensions do this: installing more than one extension at a time. As an example, I installed 4 extensions that are “recommended” by Mozilla in a brand new Firefox 3 profile. This is what I got after restarting:

That’s a pretty horrible sight even if you’re a confident add-on user. Now imagine yourself as a user trying an add-on for the first time, going to the add-ons site and installing a number of extensions that are featured on the home page. It gets worse: after you restart and see what’s in the above screenshot, close the Add-ons Manager and click OK on the modal dialog to see 3 new first-run pages opened, a wizard come up, and another modal dialog come up inside that wizard from a completely different extension but with no way of knowing it’s not related to the wizard extension:

As a novice add-on user, my first question is “how do I get rid of these windows?”, and my second question is “how do I get rid of these add-ons?”. It gets even worse. After clicking Cancel on all these dialogs, the wizard and its unrelated modal dialog and sidebar will come up every time you start Firefox until they have gotten the attention they demand. Perhaps fortunately in this scenario, novice add-on users aren’t likely to install 4 add-ons at once before restarting. Yet.

One of the most requested features for addons.mozilla.org is add-on bundles and/or shopping cart functionality. The idea behind a “shopping cart” is that you browse around the site, add add-ons to your cart that you want to try out, and do the installation all at once. Or perhaps there’s a pre-made bundle of a few of consumer-friendly add-ons available for me to try out. When all of those are installed at once, I get the above result, even as an add-on beginner.

How do we fix this?

David Rolnitzky, Basil Hashem, a few others, and myself have been discussing this for some time. We have come to the conclusion that there’s no good solution to solve this from the addons.mozilla.org/server side. Warning the user that the add-ons they’ve selected might conflict is not a good first impression for someone that’s never tried add-ons before, and may justify fears they already have about add-ons being made by third-parties. I’m making this post to ask add-on developers, especially those that are “consumer friendly” to please be as courteous as possible in the first-run experience after a user installs your extension. How do you do this?

• No first-run modal dialogs! If your extension requires an account login or other settings to be configured immediately, place these forms and elements in your first-run page, an area linked from your first-run page, or in a dialog that is only opened from a control on your first-run page. You may want to make sure your first-run page is in chrome:// if the data is sensitive or requires chrome privileges to be stored.
• Only one first-run element is needed. If you have a first-run page, you shouldn’t need to use a wizard, dialog, or sidebar too. Instead, use the first-run page to teach the user how to access the extension and allow setup of the extension right from that page.
• Standardize first-run page usage. Right now, this just means not doing anything particularly crazy or performance-degrading with your first-run page. I recently filed bug 459965 on adding a property to install.rdf that would make adding a first-run page to an extension as simple as typing in the URL to use, which will help make sure the first-run experience plays nicely for all extensions and could even be disabled if the user wanted.

If you’re an extension developer and know that your extension will make a user run away at first-run, please look into ways of improving the first-run experience. If you have a popular extension that does this, I may be contacting you soon to try working out these issues, as the success of package installs and bundling depends on a great user experience.

It’s that time again…

With the release of the first beta of Firefox 3.1 today comes everyone’s favorite release-time festivity: extension compatibility updates!

If you’re an extension developer using a maxVersion of 3.0.* or less, please test your extension before declaring 3.1b1 compatibility. Some of the changes for extension developers are listed here. 3.1b1 is an allowed version on AMO, but 3.1.* will not be added until closer to final release. Keep in mind that you can always look at the Developer Statistics Dashboard to see how many of your users are on 3.1 betas and may be marked as incompatible/disabled.

And as another reminder, if you only need to bump the maxVersion number for compatibility with 3.1b1, you do not need to submit your add-on for review. Simply use the Developer Tools area to bump the maxVersion and Firefox will check AMO for this when determining which add-ons are incompatible.

If you’re an extension user and one or more of your add-ons isn’t working in Firefox 3.1b1 yet, please be patient. Many add-ons are developed by volunteers in their spare time who may not be able to address compatibility immediately if their extension requires significant changes, which it should hopefully not. As in previous releases, Mozilla will be reaching out to developers in an effort to get as many extensions as possible updated before 3.1 is released.

AMO Developer Tools Revamp, Part 6: Manage Previews

It’s been about 4 months since I posted about the status of the AMO Developer Tools Revamp. Later this week, the beta versions of these tools currently available in production AMO will be upgraded and 3 new tools will be added. I’ll post more details about that on Friday, but wanted to highlight the features of the 3 new tools.

First up is the revamped Preview Manager:

The only specific new features with the preview manager are the ability to replace a preview’s image without losing the caption’s translations and the ability to upload multiple previews at once. The main improvement is that the management of all previews is on a single page instead of separate pages without a real connection to each other.

With this new preview manager also comes a revamp of how preview URLs across the site are formatted, which will fix the very frustrating and longstanding bug where making changes to preview images wouldn’t show up for hours.

Next up will be a post on the new Status Changer tool.

Previous post: Manage Versions and Files
Next post: Change Status

Top Ten Add-ons II

One year ago today I made a blog post listing the “top 10″ add-ons according to three different sources of data. The point of that post was to show the differences in the methods and highlight how a single add-on might be outstanding in one area (for example, downloads), but not in another (for example, user retention/active users). I wanted to show these lists for 2008 and take a look at how much those lists can change over the course of a year.

First, the data for 2008:

The first, most important conclusion I came to from analyzing this data is that the Firefox add-ons ecosystem is very healthy. It’s not stagnant, and it’s not volatile.

But total download counts can be rather old and stubborn — they encompass every download addons.mozilla.org has served for an add-on since 2004 (over 900 million of them). This gives older add-ons a pretty large advantage over newer add-ons, and once an add-on makes it into the “most popular” listing, it’s downloaded simply because it’s popular.

With that in mind, I thought I’d show 2 other “top 10″ download lists that paint a better picture of what’s happening right now.

Some things I took away from this data:

• Add-ons can make or break themselves. Video DownloadHelper is a great example as an add-on that was ranked #60 in total downloads a year ago, and has shot up to #8 - certainly not an easy feat. On the other hand, extensions like Fasterfox and Adblock (not Plus) have disappeared from every top 10 list because they haven’t been updated in around 2 years, among other things.

  Taking a look at compatibility of the add-ons above, 10 are compatible with Firefox 3, 5 support Firefox 3 and the latest 3.1 builds, and 1 (VideoDownloader) only supports Firefox 2 and is on track to disappear from the Total Downloads list soon.

• Add-on types besides extensions are gaining popularity. Looking at the 4 lists above, there are 2 themes listed: Noia 2.0 (eXtreme) and Aero Fox. But what you can’t see from the above lists — because they only include extensions and themes — is that the United States English Dictionary is actually the #7 add-on by Active Daily Users and #5 in downloads over the past year. That’s users from non-en-US locales downloading the en-US dictionary, which is very interesting in itself, but better left for someone else to analyze.
• There are popular add-ons of all ages. Of the 17 individual add-ons represented in the above 4 lists, the year of creation breakdown looks like this:
  • 2004 (5) - Download Statusbar, Noia 2.0 (eXtreme), DownThemAll!, FlashGot, Forecastfox™
  • 2005 (4) - NoScript, Greasemonkey, ColorfulTabs, IE Tab
  • 2006 (5) - Firebug, Adblock Plus, VideoDownloader, Video DownloadHelper, Fast Video Download
  • 2007 (3) - Cooliris (formerly PicLens), Fast Dial, Aero Fox (2 of these are less than a year old)
  • 2008 (0) - how many add-ons yet to be created will appear on next year’s lists?

The above points are thought-provoking and interesting, but the main thing I’m going to take away from this research is reassurance that, even with Facebook, iPhone, Ubiquity, and all of the other platforms available for developers to choose from, developing add-ons for Firefox is still useful, exciting, and relevant.