The roads I take...

Originally, I only wanted to get a few photos up fast but that ended up taking me a few hours today, esp. in adding photo descriptions, and then some OpenStreetMap updates regarding the places I've been...



In any case, I finally found the time and made selections of photos of a number of travels and events to put online (I only post selections because I often have hundreds of photos of those travels and it can probably get rather boring to view them all - but then, it takes some time to select and put up photo descriptions).



As most of those events are somehow related to Mozilla, you might be interested in some parts of them, even though the parts about conferences are usually rather small, I tend to chat more there than take photos, after all. Though, you might just like all the other pics as well...



Those galleries are newly available now:

• FOSDEM 2007, Brussels
• FOSDEM 2008, Brussels
• Canada/USA - July/August 2008
• MozCamp Barcelona 2008
• FOSDEM 2009, Brussels
• Lift09, Geneva

This list totals 337 photos, all with tags and descriptions.
(Even though this still misses MAOW Berlin 2009, from which I haven't made a selection yet, I hope I come around to that soon.)

Here's a summary of SeaMonkey/Mozilla-related work I've done in week 17/2009 (April 20 - April 26, 2009):

• Download Manager:
  The new UI patch now is available in it's final, ready-for-checkin version on the bug!
• SeaMonkey Buildbots:
  common comm-central unit test class was checked in this week and also deployed to SeaMonkey testers, removing custom factory code from the config directory.
  In other news, we have now really fixed the leaks we still had from the extension manager datasource stuff on shutdown, so I could also decrease the leak thresholds for all tests to 0 - with the only exception of Windows mochitest-plain, where we still report a 200 byte leak, probably related to some plugin stuff.
• Bug Triage and Statistics:
  Reminded on the topic of bug triage, I wrote an updated post on triage targets including the idea of changing all bugs back to UNCONFIRMED that have no comments since the new SeaMonkey project began, and in a few months go and change all UNCONFIRMED bugs to EXPIRED that haven't had a comment for a number of months. This could potentially clean up our view of the SeaMonkey product on Bugzilla a lot.
  Inspired by that triage stuff, I thought it would be nice to see some numbers of the weekly "performance" of SeaMonkey in Bugzilla, and created bug statistcs on dev.seamonkey.at as described in my recent blog post.
  There's more interesting data that could be gathered, but it looks nice that in the week I created those stats, we had 14 new bugs reported while fixing 18 and closing 30 to other resolutions by triaging. Not only did we resolve more bugs than we got new ones reported, we also fixed more than those new reports!
• German L10n:
  Once again, some string updates and cleanups were needed to keep German L10n green, nothing too complicated here though.
• Various Discussions:
  GSoC project decisions, EV cert UI, test failures, mail account autoconfig work, SeaMonkey statistics, MozCamp Wien, Mac theme rework, Linux updater issue and fixes, Mozilla 1.9.2 and Tiger support, bmo workflow, etc.

Looking into those bugzilla statistics is quite interesting: For example, in April, 68 bugs have been reported so far in the SeaMonkey product, 67 have been resolved by triage (only one less than the new ones) and 41 bugs have been fixed in that period. Esp. the latter number perfectly tells that development is moving on - given that 7 of those have been ranked with "enhancement" severity, it's clear that this active development also means new stuff coming in for SeaMonkey 2 and not just fixes of problems, even though those account for the majority of fixes.

Can you make those numbers look even better at the actual end of the month or enable us to have a good start into May by those measures? Helping is as easy as triaging UNCONFIRMED bugs and either determining that they are valid (NEW) or resolving them to INVALID, WONTFIX, DUPLICATE, WORKSFORME, or INCOMPLETE if you can't confirm them (actual resolution depends on why you can't confirm them to be valid reports, of course)!

I originally created dev.seamonkey.at as a place to have a gallery of the submissions in the process of finding a new SeaMonkey logo back in 2005.

In the recent years, I started slowly adding handy links and overviews to the main page of the site, slowly transforming it to a dashboard for SeaMonkey data that can be interesting to SeaMonkey developers (mainly stuff that's interesting for myself).

This includes Bugzilla blocking/approved/wanted/fixed radar queries for our upcoming releases or a CVS tag overview (which I needed for some time to catch when Firefox releases would be tagged), but had been enhanced over time to include the most recently filed SeaMonkey bugs listed directly on the site, current download numbers directly from the bouncer tool and a list of recent hg changesets and CVS changes that affect SeaMonkey, along with links to bug lists related to those changes.

As of last night, the main page now also features some bug statistics about the SeaMonkey product in the last 2 and the current week, with numbers and links to the following queries:

• New: Newly filed bugs in that week
• Fixed: Bugs that have been switched to RESOLVED/FIXED
• Triaged: Bugs that have been switched to RESOLVED and any other resolution than FIXED

This gives a bit of an overview of what's happening, and I hope it's useful to more people than just me. I'll happily evaluate adding other queries to those statistics when suggested here - and the same is true for adding other info on that page!

Here's a summary of SeaMonkey/Mozilla-related work I've done in week 16/2009 (April 13 - April 19, 2009):

• Release Management:
  Continuing uploads of contributed builds for SeaMonkey 1.1.16.
• Download Manager:
  Based on the new patch Callek did put up for the backend switch, I did a new version of new UI patch as well as the tests, both have the needed reviews now and only need addressing of some final nits to be actually ready for landing once the backend patch is ready for that as well.
  The progress dialogs have a first patch up and a first review pass that showed a number of additional things to work on.
  Additionally, I posted a patch for making toolkit tests pass with our UI patches, it was found to not be technically wrong but not as beautiful as expected, let's see if either Shawn as maintainer on the toolkit side or I have better ideas.
• SeaMonkey Buildbots:
  A patch for getting a comm-central unit test class in a common place was created and reviewed this week, at the time I'm writing this it even has landed!
• Linuxwochen Wien:
  I spent a good amount of time both preparing my talk and hanging around with local FLOSS people at Linuxwochen Wien.
  Along with that, I converted the core of my Mandelbrot app to a web demo and started talks for a MozCamp (or OpenWebCamp?) in Vienna in October, see my related blog post for details on all of those.
• Various Discussions:
  GSoC project decisions, fixing the workaround for the (in)famous EM RDF leak, EV cert UI, "upvar2" landing, Mozilla group/list spam, mail account autoconfig work, static builds for nightlies and releases, new SeaMonkey buildbots coming "soon"?, etc.

At Linuxwochen, I once again realized I'm way too litte networked with FLOSS people locally in Austria, but I think this time I'll really start "open source beering" around and join some regular meetups of the FSFE people, OpenStreetMap contributors or Open Source Experts interest group of the local chamber of commerce. I think the Mozilla community, SeaMonkey and I can all benefit from commucation with those groups as well as them and the open Internet can benefit the other way round. And organizing MozCamp Wien should probably help there as well.

I'd advise you to do the same: connect with your local communities and spread the word to strengthen the open Internet!

Yesterday, I held a talk on Linuxwochen Wien, a local Linux and Open Source conference. My talk was right before Brian King took the stage (who talked mostly about the add-on ecosystem), so it fitted well that my topic was "The Open Internet and Mozilla". My slides (in German) are online at kairo.mozdev.org.

I included a number of the HTML5 demos from Paul Rouget to show what open web technologies can do, but for showing the power of JIT, I showed a demo I quickly did myself:
Ripped out of the code for my Mandelbrot XULRunner app, I put up a <canvas> and some JS code up on a website, with a small HTML form for selecting coordinates, etc. - and the Mandelbrot-Web demo was done!

It's not too beautiful, just a fast hack, but running the default image calculation in both Firefox 3.0 and a current 1.9.1-based nightly, I could show how the time needed for calculation (and display in the canvas) dropped from 7.3 to 1.6 seconds on my laptop due to the TraceMonkey JIT engine.
A number of people seemed impressed with both that and the stunning video+canvas demos from Paul.

Finally, I could even announce a whole one-day event on the Open Internet right here in Vienna! As both Mozilla and the local quintessenz association (who also organized Linuxwochen) have agreed to work together in organizing this, I could point people to the MozCamp Wien taking place on October 27, 2009.
We are in the very early stages of planning, not knowing much more than the date, the topic and the format (MozCamp in the way it has been done in Utrecht in March), but stay tuned here for updates on details as we get them figured out!

Here's a summary of SeaMonkey/Mozilla-related work I've done in week 15/2009 (April 6 - April 12, 2009):

• Release Management:
  With getting good smoketesting on all major platform within two days (thanks a lot, folks!), I could finish the release process for SeaMonkey 1.1.16 and make it go public on Tuesday, fixing the recently found exploitable XSLT crash for which Firefox released 3.0.8 as well as two other critical vulnerabilities that enable us to stay secure even when Firefox 3.0.9 is being released and advisories go public with it.
• Automated tests:
  I split off the context menu test and fixes and finally could get plain mochitests and browser-chrome tests for suite/browser into the tree. Unfortunately one of the latter tests now shows a random orange, I hope to find a workaround for that.
• SeaMonkey Buildbots:
  I finally correctly fixed the crontab entries for restarting metacity on our Linux tests buildbot machine so that it shouldn't stay orange when something makes the window manager crash.
  Additionally, I added options for switching off ref-/crashtests and mochitest suites in the comm-central unit test buildbot factory so that Thunderbird will be able to share the same class in the future. (See bug 488116 about moving it to a generic place.)
• Download Manager:
  I ported the existing tests for the toolkit/mozapps UI to our new download manager UI implementation, which also made me find a few bugs I fixed in my patch, and a few things to address as followups after the main rework has landed.
  Additionally, I finished up everything I had in mind for progress dialogs, though not all reactions on my post about them sound positive so far. Some of the comments might be good to address, some my just appear out of being used to a somewhat different look (which I consider plain ugly, sorry) and might fade away when actually using the new implementation.
• Build System:
  I (hopefully) finally figured out a way to make Windows version numbers be dynamically changed based on the application's version known centrally to the build process. I hope it turns out to work correctly and be able to get into the tree soon, it will nicely improve the release process.
• SeaMonkey L10n:
  Following a face-to-face review from Axel in Berlin, I now did a new patch for SeaMonkey default profile L10n, I hope this will pass and make localizing those somewhat awkward parts easier.
• German L10n:
  Checked in the German L10n of the last set of changes from "card" to "contact" in addressbook to keep de SeaMonkey trunk complete.
• Various Discussions:
  Broken Linux nightly updates, GSoC applications, the (in)famous EM RDF leak, EV cert UI, "upvar2" landing, MozCamp Vienna, Debian packages, Mozilla group/list spam, mail account autoconfig work, QT port building, etc.

I spent all of Monday with the Mozilla Europe Community Tour, taking a trip from my home town of Steyr through the scenic "Wachau" valley along the Danube river right up to Vienna with Sonny, Greg and Arzhel, having an nice and small event with local open source folks and ending in a bar, coming home very late that night (or morning?). I met them again on Tuesday but couldn't spend so much time as I finally needed to get 1.1.16 out the door. It was nice to meet those folks and spend some time with them, if anyone else from the Mozilla community happens to take a trip to Vienna, please tell me, I always like to meet people!

And while we are on meeting people, I got word this week that my vacation in the US Gulf Region (or whatever you call an area roughly spreading from Houston to Atlanta) should actually be possible this November. If you happen to live or otherwise stay there, please tell me, would be nice to meet up!

I wrote before that I was working on the new download manager UI for SeaMonkey 2. To lift off some work from Justin's shoulders, who fights with flaky (and sometimes just-down) internet connections as well as time constraints due to multiple jobs and a very young son next to volunteering for our project, I decided to also do the redesign of the per-download progress dialogs that we still want to and will support with SeaMonkey 2.


Using those dialogs instead of the download manager window is not the default, and I've never used them since the suite had the download manager UI, but here's how such a dialog looks on SeaMonkey 1.x (on my Linux machine, this is from a German L10n build):



This UI had a few problems from my POV: The target and source locations grow out of the visible area, there's a lot of white space wasted, many buttons that you need to parse to know what they're doing, and it generally feels a bit old-style to me (not to talk about the code, that really calls for throwing away and rewriting). Because of that, I decided to rewrite the whole UI, based on what toolkit's download manager is showing for its entries, only taking what information should be available from the old implementation but redoing everything else. Yesterday, I basically finished my implementation and while I'm holding back on attaching the code to the bug report for the moment to make it base on a hopefully-soon-appearing new backend patch from Justin, I requested UI-review based on screen shots.

This is how the new dialog looks while pausing or downloading a file:


The dropdowns you see on the local file name and the download host come up on either left or right click on those labels in the dialog, and the download can be controlled with the pause/play/cancel icons near the left browser of the dialog. Yes, it's not a copy of the old UI in new code, it's a complete rewrite of the functionality, trying to make the dialog cleaner and easier to understand. What do you think?

And for those of you who only think "Eww! Download progress dialogs are soo backwards, only really old software uses such a thing!" you can be assured that the default mode of operation for SeaMonkey 2 will be to show the download manager window instead (a hidden pref will probably even allow you to switch to the toolkit implementation), which will look like this:

Reading David Dahl's post on "antisocial networking", I finally found a way to express what's bothering me with today's "Social Web" (no, I don't call it "Web 2.0") services:

People, and even the services themselves, talk about them as "Social Networking" but all I see is monolithic, non-networked single-corporate-hosted blobs.

Where is the network of independently operated Facebook servers that are all can easily communicate with each other but are not in control of one giant operator that has an unclear objective?

Where are the same networks for twittering, sharing personal map data (think "Google My Maps"), videos, photos?

Where are the messaging services that connect social networking messages with instant messages and email?

If those services were on independently operated but interoperable services like blogs or the original web, there wouldn't be a few large operators in charge of all datamining, we wouldn't need to fear that one service goes down because its operator goes bankrupt or some parts of the service go down because it's sold or even one company going towards a data monopoly by acquiring one social web service after the other (see Google buying twitter, etc.).

I want control of my data (e.g. by hosting a good amount of things myself) but still have all the cool tagging stuff and other awesomeness out there being connected with it. But all I see is monolithic services instead of real social networks.

No, I don't have a Facebook, Twitter or Flickr account - yet. Even in 2009. The thoughts above are probably the most rational expression of some part of the uncertain feelings that drive me to still appear so much "backwards". I really care about the open Internet, but that includes open networking, and I still have some problems seeing that in those services.

What about you?

Here's a summary of SeaMonkey/Mozilla-related work I've done in week 14/2009 (March 30 - April 5, 2009):

• Release Management:
  I started the release process for a 1.1.16 release that fixes the recently found exploitable XSLT crash for which Firefox released 3.0.8 as well as two other critical vulnerabilities that enable us to stay OK from a security POV even when Firefox 3.0.9 is being released and advisories go public with it.
• SeaMonkey Buildbots:
  For getting a fix for the SeaMonkey MacOS mochitest breakage to actually work, I needed to switch our buildbot configurations to calling the make targets instead of runtests.py directly, and so I decided to give the buildbot configs a long overdue overhaul to be more similar to what the Firefox buildbots are calling. Along with that, I enabled a11y tests for SeaMonkey.
  I found one more abstraction problem that keeps us from going all the way with both builders and testers (and make both use generic pools of slaves), but I spent lots of time this week one making the tester buildbots use a configuration that is very much like what the Firefox buildbots are using. The builders will follow when the abstraction works properly and ultimately we will merge both together and even enable release automation with the same master and the then-existing generic slave pools.
• Automated tests:
  I did a new patch for browser mochitests, but correcting those context menu accesskeys along with that isn't that easy after all.
• German L10n:
  Updated German strings to current SeaMonkey code.
• Various Discussions:
  Modern theme updates, GSoC applications, www.mozilla.org redesign, mailnews disentanglement, security firedrill, stringbundle implementations and getting formatted strings, BMO process changes, etc.

The Mozilla Europe Community Tour makes stopped in my home town of Steyr on Sunday and we had quite some fun then and on Monday. If you are somewhere along their way, feel free to contact Sonny, Greg and Arzhel, they're cool guys and surely happy to meet you!

Following the security "firedrill" release of Firefox 3.0.8 last weekend, we'll also be updating SeaMonkey 1.1 for the XSL Transformation vulnerability that caused that Firefox update, as well as two other critical security vulnerabilities that can only be disclosed once Firefox 3.0.9 is being released.

Please help testing our SeaMonkey 1.1.16 candidate builds - preferably via performing the smoketests from the, please go to and do a SeaMonkey 1.1 test run on Litmus.

The sooner we have verified that the builds still work well with those fixes, we can ship the more secure version to all our users!

As the SeaMonkey Change-arounds homepage says:
Get the coolest Change-arounds for SeaMonkey 2 from there today, while they're still hot!


(Too bad that localizations haven't caught up with that change-around in time...)