The roads I take...

You might know Air Mozilla, the Internet broadcast show of Mozilla, featuring "live and pre-recorded shows, interviews, news snippets, tutorial videos, and features about the Mozilla community" (quote from their website).
The first two shows, which are available for viewing and for download from their website featured interesting live interviews of "Chief Lizard Wrangler" Mitchell Baker and other Mozilla Corporation people in Mountain View as well as clips from other Mozilla Corporation offices (e.g. in Canada and Japan) and from the Firefox Flicks contest.

While very interesting, this was, probably as expected, very much about Firefox and/or the Mozilla Corporation. But this will change - at least somewhat: A few weeks ago I was contacted by the main producer of the media content, who happens to be a native Austrian living and working in the San Francisco Bay Area. And today, as he is here for visiting his old home country, he stopped by for an interview about SeaMonkey!

We tried to find some place typical for Vienna to record at, but unfortunately all those cool places like Schönbrunn with its parks are bad for recording a video interview when it's raining, so we decided to go for a typical indoor location, like, say, a Café (you know, Vienna's known for its Cafés and apparently this place was the first to be known for mixing milk and sugar into people's coffee). And so we ended up in... a Starbucks. Real Austrian heritage. But then, this is about an international project with a strong center in the US, right?

I tried to tell something interesting about SeaMonkey - I hope I didn't mess up too much... See and hear for yourself in one of the next Air Mozilla episodes!

Here's a summary of SeaMonkey/Mozilla-related items I still did come around to work on in week 34/2007 (August 20 - 26):

• Website work:
  After finally Mozilla suite users got a reworked start page that tells them their product is outdated and they should upgrade (to Firefox and Thunderbird, SeaMonkey is only mentioned in a small paragraph at the bottom and only visible after scrolling on most screens), some SeaMonkey users complained they got to see that page as they are still using profiles they had in use with the suite and never changed this start page.
  If fixed this with a JS redirect to the SeaMonkey start page.
  
  That reminded me of an issue I had in mind for a long time, and I worked out a proposal for a new SeaMonkey start page that should be nicer and more helpful to most users.
  
  Finally, I picked up the plans to create a separate SeaMonkey project website and spend days of work only on this.
  The current state is now available for preview as well (this is still a work in progress).
  
  I'd be happy about feedback - please post this in the development newsgroup rather than here on the blog though.
• Code Reviews:
  I reviewed an interesting way of fixing accesskeys in error console and one more patch for the debugQA extension.
• Various Discussions:
  PrintUtils, issues and joy surrounding venkman/chatzilla and their maintainers, notification bars, XUL vs. HTML, www.mozilla.org future, SeaMonkey window icons, etc.

The website work stole almost all my time, so I was a bit slow on all kinds of other things. Sorry if I may not have come around to do some other things - I hope I'll be able to come back to those once the new website nears even more a state with which I as well as other community members will be satisfied.

While the SeaMonkey project itself has proven itself as a mature community that can do quite well, and we even get better and better with our migration to the "new toolkit", some code parts lag a bit behind, like those targeted by my bug bounty program. But there are also other areas that need help, one of which is artwork.

There's a long-standing open bug report around on one area where we badly would need someone to help us out: we need new icons for all our application windows.

This might sound like an easy task, but there's quite some rules to follow, which makes this hard work - and we need a bit over a dozen different icons!
Those are the requirements we have for those icon designs:

• We need to replace all icons we have that contain "window" in their name, see our Windows icon collection for what those are and their current versions (still those made a long time ago for the Mozilla suite).
• The new icons need to make clear at the first glance that they are SeaMonkey icons, so they should visually refer to the SeaMonkey logo in some way.
• Similar to that, the icons need to be consistent enough to each other in style and basic look, so that a user sees at the first glance that all those windows belong together.
• At the same time, the icons need to be distinct enough that a user does not mistake one of them for another and they need to clearly enough identify the purpose of the window they label.
• The icon designs also need to fit in with the new icons in the reworked default theme of SeaMonkey 2 (trunk nightlies), so a look similar to those new icons is preferred.
• We need all those icons in different size and color depth variations, at least 16x16, 32x32, 48x48 and 128x128 with 24bit RGB and 8bit alpha (i.e. 32bit RGBA) as well as 1-bit-transparent versions with 24bit RGB colors, 256 colors (custom palette) and 16 colors (standard EGA palette), all for the 16x16, 32x32 and 48x48 sizes.
• We need to get all those single images in PNG format, additionally providing a Windows .ico would be nice but is not strictly required. We need to create icon versions that fit for Windows, OS/2, Mac and GTK2, but once we have the raw images to use for those, it's just technical work to produce the correct file formats from them.
• Due to size constraints, it might be a good idea to have minimalized shapes for the 16x16 versions that might look a bit different than the respective bigger icon sizes, just like the current icon set (see above link) has. That current set is a good example for everything we need, actually - just that it still is based on the old Mozilla suite "m" logo and the old "Classic" theme imagery, which both are gone in SeaMonkey 2.
• Last but not least, all the copyrights for all imagery must be assigned to the SeaMonkey Project / Mozilla Foundation (per at least an email from the creator to the Mozilla licensing group and the SeaMonkey Council - Mozilla may additionally require you to send a signed assignment message per paper mail), we'll then set them under the MPL/GPL/LGPL tri-license with committing them to the source code repository.

If those requirements don't intimidate you and you feel you have enough artistic talent and icon design knowledge as well as want to contribute some work to the SeaMonkey project, then please create some preview designs (a selection of icons, including at least browser and mail in 16x16 and 32x32 versions), post them as an image somewhere publicly on the web and inform us in the SeaMonkey development newsgroup/mailing list!
Once we arrive at a design we all like, it'll be time to work on the full set and go to the bug report, but before that, the newsgroup is a better idea.

Oh, and if you need a SVG version of our logo to work on such icon designs, please contact me at kairo-at-kairo-dot-at or on IRC - we're still waiting on getting a trademark policy up before we will publicly put up such a SVG image.

I know the target for such icons is pretty high, but I hope we'll get someone to provide them and vastly improve the look of SeaMonkey 2

This status update is rather short, as I said previously, I spent Monday to Wednesday at home and mostly away from computers (see IRL://at/home/ ).
Additionally, I worked a bit on my web CMS/community system, automatically creating web-based documentation from source code comments - I hope that'll make further development work easier.

Here's a summary of SeaMonkey/Mozilla-related items I still did come around to work on in week 33/2007 (August 13 - 19):

• Places History:
  As mentioned here a few days ago, I checked in the possibility build the new "places" backend for history in SeaMonkey with a special configure flag. As the UI doesn't work completely with this yet, we can turn this on by default yet.
• Code Reviews:
  I reviewed a small string addition to help Lightning integration. Additionally, I swallowed the (for me) bitter pill and read through python code to review improvements of the new compare-locales script so that it works with more than just Firefox and Thunderbird.
  I also looked into reviewing the patches to get profile switching to work again from a running SeaMonkey, but I think we still have a few open items to discuss there, even if the patches all meet the main objective of the bug.
• Infrastructure (tinderbox, AUS, website)
  As the Mozilla project converts to new Windows build machines built upon MozillaBuild (MSYS) and VC8 SP1, I filed a bug to also move community machines, including the SeaMonkey one, to this reference platform.
  A bug for switching to new Linux images does exist already, and I hope hope we'll get a box for L10n builds as well.
  On a different front, a machine for automatic updates of community products has been set up, once everything works for Sunbird, we'll look into getting it ready for SeaMonkey as well.
  After a long time, I've once again looked into setting up seamonkey-project.org as a separate website domain (when we do this, hosting will still be done by Mozilla, we need no other offers there). I still need to really get familiar with the website build system we are provided with and find out what kind of design would fit us best, but I have "something" working locally for testing. www.mozilla.org still works fine for us basically, though, so there's no real pressure on that, but at some point we'll probably arrive there.
• L20n:
  While that proposed new framework for localization is cool, there's not much discussion or work going on around it, which is sad. I tried to get at least discussions going again this week by posting to newsgroups, my blog and mailing a few peers.
• Various Discussions:
  IPC, venkman chrome errors and making it an extension, killing usage of old chromereg compatibility, killing old chromereg, suite start page changes, etc.

In other news, I did finally find the time to request a new passport and should get it this week, so that I'll even be able to travel to the US again. I hope I can visit the Mozilla headquarters at Mountain View some time in the near future.

It's been a while since I blogged about the L20n framework - and unfortunately, not much is happening in that area at the moment.

As I think we badly need a better framework for localization, I just posted to Mozilla newsgroups about the framework and called for comments on it.

Many people were left in a bit of confusion about the proposed framework and file format, and I think we should clear that up, get the discussion going and arrive at a real spec, a prototype implementation and maybe even a prototype of an L20n tool for localizers.

We need to know what developers, localizers and L10n tool authors think about what Axel proposed and how we can improve the proposal even further.

Please also add your comments on this proposed framework in the newsgroups!

I've just checked in the first step for using places history in SeaMonkey.

"Places" is the new SQLite-Database-backed scheme for saving URLs of any interest which is built into the Mozilla toolkit and is already used by Firefox (for browsing history, current Firefox3-preview-nighties also use it for bookmarks). Our target is to use this component at least for history in SeaMonkey 2 instead of the mork-backed xpfe history we still have in use for now.

The current patch makes it possible to build SeaMonkey with the places backend with the --enable-places configure option. URLs from old mork history will be automatically imported into the new places.sqlite file in the user profile, visited links and the urlbar autocomplete popup work without any further work, but the history window and sidebar stay empty, as they are based on RDF templates which don't work with SQLite/mozStorage. mozStorage-based XUL templates should make it easier to get them working again, once those make it into the toolkit code.

I hope some people who are doing builds themselves will enable the new backend and do some testing despite those UI bits not working so that automigration, urlbar autocomplete etc. get their share of testing before we fully switch to the new scheme.

Sorry for being very late with my status update from last week this time. I've taken some days more or less off from work from Saturday to Wednesday evening and spent them at home with my parents, in municipality and bank offices to get a new passport and get my finances better in order, and at my brother's 25th birthday party.

Still, better late than never, here's my summary of SeaMonkey/Mozilla-related work I did in week 32/2007 (August 6 - 12):

• Releases:
  Uploaded some more builds for SeaMonkey 1.1.4, thanks to all contributors providing those!
• German L10n:
  I went into the next incarnations of the security and dom L10n sync patches, I think we're nearing a usable state of both.
  We're currently trying to get a newsgroup for German L10n, I hope it'll be created very soon.
• Source L10n:
  I filed a bug to get a machine for doing langpacks and L10n repackaging for SeaMonkey, which currently is the major blocker to get locales into CVS in a testing mode.
• openURL():
  As this sounded like something that should have happened for a long time, I moved Philip's relatively generic function to open URLs from any Mozilla-based app from extensions.js to contentAreaUtils.js, so that it can be used by other consumers than Extension Manager.
  I hope such other consumers will be added soon.
• KaiRo.at Bug Bounty Program:
  Finally could disclose this project and announced it on my blog.
  Even in the first days, some progress could be seen on the first items on the program's list.
• Small work:
  I finally got my image region reftest into the tree that I've blogged on some weeks ago.
  Also unbitrotted the places history switch patch which will allow people doing builds themselves to easily test the new history backend in SeaMonkey.
• Various Discussions:
  Current and future build tools, automatic update system,etc.

I hope to provide this week's status update on-time again, lately it's becoming somewhat of a bad habit for me to provide those updates late, I'm trying to not keep this one.

It was nice to see a first patch for an item of the Bug Bounty Program just one day after announcing this program!

Daniel Brooks has picked up the page info improvements, and his newest patch looks really good. We're getting a reworked, more polished page info window with more functionality (without losing any compared to the old one, the forms and link tabs are still there) - oh, and this is probably the first part of SeaMonkey that will see some feeds support as well (by listing feeds available on the page).

Just a day and a half later, Teune van Steeg attached a first patch on another bug from this list, adding browser notifications / info bars to our software. This is an important feature for use with other code, and I'm happy to see something moving there as well.

Thanks to both contributors for picking up those items!

While nobody has started to work on it yet, I had some talk about feeds support on IRC: Robert Sayre (sayrer), who worked on toolkit's feed parser (which we should use), is willing to help anybody working on our code when questions arise, Myk Melez (myk), who did the Forumzilla extension, is willing to answer such questions as well. As there are two different ways of integrating feeds with mailnews out there (Thunderbird's and Forumzilla's), we got into some talk about that, but both our SeaMonkey mail owner Karsten Düsterloh and me think the Thunderbird solution of having them in special accounts is the cleaner solution, which we prefer.

I hope someone pick up this and the other items on the list soon, so that we see even more good progress for a better SeaMonkey 2.

I've been thinking about this project for some time, and working on it behind the scenes for a few weeks now - finally it's ready to be announced publicly:

The KaiRo.at Bug Bounty Program is being launched today!

This program, launched by KaiRo.at - Robert Kaiser IT-Services, aims to "encourage developers to get involved with SeaMonkey by rewarding them with some hard dollars for contributing important code to the SeaMonkey project", as stated on the program's pages.

The program is currently awarding a total of 2700 USD to work on a list of 7 critical bugs/features on the path to SeaMonkey 2, including dynamic UA spoofing and feed reading, as well as download manager, Lightning integration, page info, plugin finder and browser notifications. That's the current set of bug bounties for 2007, there's still a slight possibility that KaiRo.at may even extend this program though as we see fit. Any developer completing the work as described on the program's pages is eligible for the bug bounty assigned to the specific task.

I hope those bug bounties will be an additional motivating factor for people to work on those features for SeaMonkey, but also hope they will continue to stay with the project after completing those immediate tasks.

And here's a short Q&A about this program, to sum it up once again:

• Who is KaiRo.at, what is their interest in this and where is this money coming from?
  OK, that's 3 question at once, but the answers cumulate for the most part: KaiRo.at - Robert Kaiser IT-Services is the one-man-business of me, SeaMonkey Council member Robert Kaiser. The money comes from my earnings through Google AdSense on the SeaMonkey German website - and as that income rises with SeaMonkey adoption in German-speaking countries and my business is mainly around SeaMonkey, my interest is in improving SeaMonkey so that more people will use it. Sounds too good to be true? Well, it isn't. It is the whole truth. Believe me.
• What tasks will be awarded with which amounts of money?
  Look at the bug bounty list for 2007 for which tasks are currently in the program, and the detail pages linked there for the amount and detailed description of the work to do there.
• Who will get the money?
  As stated on the main program page, the developer who implements the vast majority of the task described in the detailed per-bug work description page will be awarded with the sum stated in that latter page. If multiple developers worked on the task or something else out of the ordinary happens, the ultimate decision is up to KaiRo.at (Robert Kaiser).
• What is the motivation behind the program?
  What's driving this program is the idea of rewarding people for their work on important parts of SeaMonkey and to give an additional motivation for developers to implement the listed features. The money is earned through the community using SeaMonkey, it's only fair to give something back to those who work on improving this software.

With that, happy hacking on an improved and much better than ever SeaMonkey 2 codebase!

I could get done some interesting things done this week - here's a summary of SeaMonkey/Mozilla-related issues I worked on in week 31/2007 (July 30 - August 5):

• Releases:
  Your probably read about it, we managed to fix a few nasty security vulnerabilities with our SeaMonkey 1.1.4 release, I spent lots of time on that - with great help from the community!
• Passwordmanager conversion:
  We had lots of talk about this - we want to turn to toolkit's login manager, but mailnews still needs wallet, mainly because login manager misses an interface it needs.
  Because the wallet password manager and login manager listen to the same preference, we've had two prompts for saving passwords come up on logging into websites, so we needed to disable building most of login manager again for the time being.
• Themes and RTL support:
  Along with the wallet replacement work, the issue of RTL support came up in the form of browser autocomplete and we came into a more general RTL discussion along with that. It would be nice to support RTL languages better in SeaMonkey, which will need some modifications to the remaining parts of the Classic theme - ideally even to Modern, but Classic comes first. Any help on that would be appreciated!
  Additionally, I once again did update my EarlyBlue theme to work with the current trunk of SeaMonkey.
• Cleanup of old xpfe code:
  Steffen Wilberg has noted that xpinstall/standalone had been abandoned for a few years already, so I removed it from the CVS repository.
• German L10n:
  Finally got the editor L10n sync checked in, as well as some security string updates in core triggered by our discussion for German.
• Lightning Calendar:
  As I mentioned earlier, I turned on Lightning for SeaMonkey this week. Be sure to read my blog post about it if you're interested in that topic.
• Statistics:
  I worked on and posted a few statistics here on the blog, I hope they gave some good insights about what is going on
• (Undisclosed Project):
  After some further work on it, my SeaMonkey-related business project is ready for an announcement, which will follow this post within hours.
• Various Discussions:
  Current and future build tools, FF3 download manager UI, automatic update system, Vista compatibility, etc.

I tied to increase the frequency of blog posts to not only have status reports - I hope this is helpful even in the holiday season. Of course, one never knows about that...

We have just released SeaMonkey 1.1.4, our most recent security release that fixes a few very nasty security vulnerabilities.
If you're still running any older version of SeaMonkey, or even a Mozilla or Netscape suite, get this release now to be sure to not run knowingly insecure code.

And if you're using one of our main 3 links to our official release builds, i.e. those that go to download.mozilla.org, your download gets counted by our download redirect tool called "bouncer".

As already stated previously, downloads of any other builds (.zip or tarballs, other platforms and other languages) are not counted, as well as downloads that are issued directly from FTP servers, or through other means of distribution (Linux distro packages, etc.) - so the real number of SeaMonkey downloads is in the dark to us, but the counts of the three main download links should cover a significant portion of our downloads.

And usually, when I poke the Mozilla build team to add new download links for a release to the bouncer tool, I also ask them for the current numbers.
As of today, when we added the 1.1.4 release, the counted downloads for SeaMonkey versions from 1.0 to 1.1.3 sum up to over 1.8 million downloads already!

The graph below shows a per-version split of those (blue bars and left scale):



1.1.1 ranks highest with 386k downloads - but then, this version also had the longest lifetime (together with 1.0.8) of all releases, it held the title of being the most up-to-date SeaMonkey version for 91 days until 1.1.2 was released.

So I figured ranking the releases might more sense when taking the lifetime of those releases into account - the short 15-day-only timespan until 1.1.3 was followed up by 1.1.4 makes this even more important. What I did was diving the total downloads per release by the days in which that release was the most current one (on its branch) - in the graph above, this is the red line, using the scale on the right side.

The outcome is quite interesting: We started off around 1800 downloads per day for 1.0 and 1.0.1, with a spike to slightly over 3000 for 1.0.2 - but people obviously didn't want to adopt 1.0.3 last summer, it had only about 830 downloads a day. After that historic low, we saw a steady increase of download rates, nearing 5000 downloads/day lately, and 1.1.3 even had 5880 downloads on average in the 15 days it was out. Of course, the alpha/beta releases for 1.1 and those 1.0.x releases that only continued that branch while 1.1 was already out the door are way below the numbers for the most-current releases at any time.

And one last interesting number: In the 550 days since SeaMonkey 1.0 was released, bouncer recorded an average of 3376 SeaMonkey downloads every day!

I think we are on a good way and especially the high download rates for all 1.1.x releases are very encouraging.

You might have heard about the recent Firefox 2.0.0.6 and Thunderbird 2.0.0.6 releases and from what vulnerabilities they state, adding the nasty mailto: problem of SeaMonkey itself, you might have guessed that another release of our suite might follow shortly.

The truth is that we just got the fix for that last vulnerability into the tree yesterday, I did tag the tree for 1.1.4 immediately, spin the builds over night, and we are into testing today.

Why I'm posting right now is that this evening, we got two complete smoketest runs on Windows and one on Linux already, all three by German community members; and in the directories I have been readying for the release on Mozilla's FTP staging server, we have - next to the official English Windows, MacOS X and Linux i686 builds - already contributed builds for Linux x86_64 (by an Australian community member) and OS/2 (by another German community member) as well as my German langpack, win32 zipfile and Linux tarball - and win32 installer, contributed by yet another German community member!

It's really, really great to see this great worldwide community at work, and sometimes it's overwhelming how active the German-speaking part of it is.
And that's all in addition to our developers from the United Kingdom, the USA, once again Germany, Sweden, and lots of other countries, the localizers that allow us to have SeaMonkey available in 17 languages at the moment (with even more to come), people working on extensions, reporting bugs, and of course the high number of users.

A big "Danke!" to everyone who makes all this happening. It's so cool to be part of this community - and you all are the cause of this great feeling I get when sometimes sitting back and looking at everything that's going on here.
Thanks once again. You rock!

As I mentioned here earlier, the default start page for the German versions of SeaMonkey and the Mozilla suite are running from my site, and I have pretty good statistics about relative usage of those application suites in the German market because of that.
While in February I had still almost 2/3 of the hits coming from Mozilla suite and a bit short of 30% from SeaMonkey, in April Mozilla was down to about 55% and SeaMonkey around 37% - and that trend did continue.

Here's a current graph of hit percentages by browsers:


And you'll see from this that it's time to celebrate - SeaMonkey has finally overtaken Mozilla in July!
The July numbers of hit percentage are SeaMonkey leading with 46%, followed by Mozilla with 45%, Firefox is at 4% and MSIE at 1.5% - all others are well below 1% of total hits on the SeaMonkey German site.

A breakdown by versions is also interesting: SeaMonkey 1.1.2 leads with 15.9% followed by 1.1.1 with 15.0%, the strongest Mozilla version is 1.7.12 at 10.6%, followed by 1.7.13 at 6.3%, 1.7.3 at 5.6%, and 1.7.5 at 3.9%. The newest SeaMonkey version, just release on July 19th, is at 3.6% in July's totals already, Mozilla 1.6, 1.7.11, 1.7, SeaMonkey 1.1 and Mozilla 1.7.8 are all still above 2%, all SeaMonkey 1.0.x versions sum up to about 9% of all hits - that's practically all versions of 1.0.7 and below, the 1.0.8 and 1.0.9 which came out post-1.1 are hardly noticeable in the statistics.

This all gives me the impression that the update notification of SeaMonkey surely works better than no notification at all like we had with Mozilla back then. Also, people who stick with some version seem to not even step up to the respective conservative security updates, like the 1.0.x splitting shows. Many people seem to follow the updates we ship though, when they get notified of them, as the decent percentage of 1.1.2 and 1.1.3 shows, but they might not follow them immediately, as the still high 1.1.1 number makes clear.
I hope our move in SeaMonkey 2 to the automatic update system (AUS) also used by Firefox will make people follow the security updates even more closely.

Still, a Gecko percentage of 96% always sounds nice on a site with over 3 million hits per month - with 2% of the rest being bots, 1.5% Trident (MSIE), 0.15% Presto/Opera and 0.08% KHTML (Safari/Konqueror), just if you might wonder where those other competitors are. Sure, this is a Site dedicated to Gecko browsers and holding the default start page of German SeaMonkey and Mozilla, so Gecko domination is of course expected.

As we're already on this topic and we already know that those hits ale almost all from SeaMonkey and Mozilla suite users, let's take a look at the most-used OSes as well:

1. Windows XP - 78.50%
2. Windows 2000 - 8.95%
3. Windows 98 - 2.49%
4. Linux - 2.28%
5. Windows Vista - 1.73%
6. MacOS X - 1.47%
7. Windows ME - 0.97%
8. Windows 2003 - 0.61%
9. Windows NT 4.0 - 0.38%
10. Windows 95 - 0.09%
11. Solaris - 0.03%
12. OS/2 - 0.01%

I've taken the liberty to only use version splits for Windows, and take the sums for others, as well as exclude everything below 0.01% of usage (i.e. below 300 hits this month). XP clearly dominates suite installations by far, but I wondered how low Vista usage seems to be, and how high Win2000 ranks there. This high usage of Win2k likely points to many corporate and university etc. installs being out there.
Another interesting point is that Windows versions below Win2k, which will be dropped by Gecko 1.9 and therefore also SeaMonkey 2, are already below 4% of our users. Given that SeaMonkey 1.1.x will be supported until April or May 2008 at least (6 months after Gecko 1.9 is released with Firefox 3), there's a good chance that this percentage will decrease enough that not too many people will get hurt by this (esp. as I don't know hoe Mozilla and SeaMonkey users are spread across those OSes, but expect that older OSes have potentially higher usage of the older product).

The trend of SeaMonkey adoption and Mozilla suite being abandoned continues well, but probably slower than some people might expect - for me, at least, it can't go fast enough