The roads I take...

Another interesting week has passed, and here's a summary of my SeaMonkey/Mozilla-related work items in week 43/2007 (October 22 - 28):

• SeaMonkey Releases:
  I spent some time uploading additional builds for SeaMonkey 1.1.5, and started discussion on 1.1.6, which we probably need to do fast for layout regressions that unfortunately crept into Gecko 1.8.1.8 - this is one reason why we'd need better testing of candidate builds so such regressions can be fixed before instead of after a release!
• SeaMonkey 2 Planning:
  I blogged on my view on SeaMonkey 2 goals this week, I'm also planning on writing up a similar doc for the whole SeaMonkey project soon. Some progress is happening again on a few major open areas for the future of the suite, like preferences rework, feed reader, IPC or mozStorage queries. Additionally, I had an interesting IRC talk about user agent stuff, maybe Flock people are interested in the same dynamic spoofing and evang feature we'd like to see, even Prism could possibly want this (Mozilla Labs are not using a UA containing "Firefox" for this project).
  I think we should start assembling criteria for what we need to release an Alpha of SeaMonkey 2 code, as I think we are coming to a point where such a state might be getting within the range of vision.
• Source L10n:
  I suck. Everything (!) depends (!) on me (!) making an announcement in mozilla.dev.l10n so that localizers can start with SeaMonkey trunk L10n in CVS, but I didn't yet come up with the time to write it. I hope I can do so soon, there are a few things I need to point out there before people can import their stuff.
  On other fronts, the patch for ChatZilla as real extension in SeaMonkey went in, so I can continue to work on CVS-based ChatZilla langpacks.
• German L10n:
  I'm continuing to try to keep trunk as well-working as possible, another set of reviews is pending at the moment.
• Various Discussions:
  ChatZilla-as-extension, release strategies, prefwindow migration, browser notifications, themes and icons, mozilla2, etc.

Currently, big things are going on in many areas regarding the future of the suite, I hope I can report many of them progressing to important milestones or even being completed in the next weeks!

Some of you might know that the SeaMonkey team is working hard on something called "SeaMonkey 2" (sometimes dubbed "suiterunner") and of which some pre-alpha testing versions are floating around. Every now and then questions come in about something someone has heard about this magic piece of software that we believe to solve all problems of this world. OK, OK, yes, I'm exaggerating here: not the whole world, probably not even all problems of SeaMonkey 1.x - but we're trying to fix as many problems as possible, and surely some very big improvements will be made in that version.

If that was the only goal for us, we probably would never be able to release this "SeaMonkey 2", as there's always room for further improvements. So, I'm trying to summarize my view of what goals we want or need to meet for that big milestone in the project:

• Modernize the backend framework. This step, also known as "switch from xpfe to toolkit", brings us to building upon even more of the same base shared with Firefox, Thunderbird, Sunbird, Songbird, Miro and other Mozilla-based software. This also implies adopting the much better extension management those applications already have as well as a list of other good improvements.
• Reduce and consolidate SeaMonkey-specific code. The more code we can share with others and don't need to maintain completely within our small SeaMonkey team, the better. Because of that, we try to reuse as much code as possible from the core framework of Mozilla. At the same time, we're moving the code specific to SeaMonkey but previously spread across multiple paces into a common suite/ directory in the Mozilla code tree, so that maintenance of that code becomes easier.
• Modernize the look while still keeping the typical suite feeling. We're replacing the aged icon set we inherited from the Mozilla suite and in many cases even from the original Netscape Communicator with a set of icons that fit in better with the current breed of OS desktops. We might do other small updates to SeaMonkey's look as well, along with that. At the same time, we're trying very hard to keep the typical suite feeling alive by making the new pieces consistent across all parts of the application and leaving all typical and familiar elements in their usual places and keeping the functionality intact that our users have grown accustomed to. That way, our goal is to carry the suite concept from the 1990's to modern work environments.
• Improve functionality without bloating the UI. We might add feed functionality, new Add-Ons management, an improved download backend, reworked page info, new storages for cookies, history, maybe even bookmarks, and even more things, as far as we have people able to do the work in it. In all those cases, the UI needs to integrated without complicating or vastly changing our existing user interface. Though SeaMonkey is targeting advanced users and web developers, who can (and often want to) handle our extensive menu and graphical pref systems, we need to be careful to not make using our software even more complicated. While I can for example imagine adding a find-as-you-type search bar to history or bookmark windows, I don't think adding every imaginable thought-to-be-cool feature to that system while merging history, bookmark and downloads into a single overloaded organizer window is a way we should go. Other apps might be trying to do that, we'll see how it works out. If adding functionality to enable extensions to do that with our backend doesn't cost much overhead, we should look into that instead.
• Improve extensibility. One often-cited reason for Firefox' worldwide success is that it's easy to extend with a growing list of add-ons. While installing extensions in SeaMonkey 1.x works fine, SeaMonkey 2 will pick up the improved system with good integrated uninstall, disable and upgrade capabilities. To further improve on that, we should try to provide similar or the same APIs to extension authors as applications that are more widespread and compatible, so that making existing extensions work with SeaMonkey will be even easier. That means e.g. exposing the Firefox sidebar APIs for the SeaMonkey sidebar (without copying the look) or adopting the same element IDs as other apps in XUL where it makes sense that one extension overlay can hook into our UI with the same code as for the other app.
• Improve customization, keep useful defaults. We want to enable users to customize for example the toolbar layout of the browser, so they can e.g. move the home button to the main toolbar or maybe add elements not yet present in our default UI, but the defaults should stay unchanged to SeaMonkey 1.x unless there are very major reasons to change something (e.g. security or clearly major usability improvements). Changes to defaults should be handled conservatively, while changes to customization should be handled progressively as long as they don't reduce usability or unnecessarily increase complexity.
• Integrate better with current OS desktops. This point is especially important in the case of Windows Vista, where SeaMonkey 1.x is known to have problems e.g. with getting registered as the default browser or mail client. This should get fixed in SeaMonkey 2, as well as getting elevated privileges where needed for installation purposes, etc.
• Enable automatic partial updates instead of full re-installs. With SeaMonkey 1.x, one needs to download a new version and manually reinstall it to get the newest security updates. In SeaMonkey 2, we want to adopt a system that can automatically download only the needed changes and apply them to an existing installation, so that updates are smoother and will be done more eagerly by our users.
• Automatically build and release localizations. SeaMonkey 1.x versions in languages other than US English get contributed by the localization teams themselves and are only available as those have time to rebuild them and only in the versions our volunteer localizers can provide. When doing a SeaMonkey 2 release, we want to be able to build localized versions right along with the original builds and release them in all available languages and for all three primary platforms at the same time, includes the previously mentioned automated updates.

As huge as some of those goals might look, we've already managed to go a major step of the way for most of them, and I'm pretty confident we can manage to meet the vast majority of them almost completely. If you're interested in a target date for releasing this major overhaul of our all-in-one internet suite, the only thing I can tell you right now is "when we are satisfied we have met all the gaols we have in mind for SeaMonkey 2". Oh, and that it won't be sooner than the first stable release of Gecko 1.9, which will be the Firefox 3 release. We need this stable base to build upon, so we at least need to wait for that. I still think our release will once again lag behind the major Firefox release, but we'll see how fast we will be in fulfilling the goals we have set up for ourselves.

We'll probably do an Alpha when we're far enough on all the tasks that we can encourage testing by a broad audience, and we'll go into beta when have the feature set completed and only stabilizing and cleanup work is left. Until then, those of you who dare to possibly test unstable, not completely working code and have good backups of their private data so that possible damage to it is no problem, can test our trunk nightly builds. They meet a good subset of the goals above already, but are very experimental and not for the faint of the heart - esp. not for daily use by normal users - yet. Still, as you can guess from the list in this post, we're working on creating a pretty exciting piece of software here.

Here's a summary of SeaMonkey/Mozilla-related work items I spent time on in week 42/2007 (October 15 - 21):

• SeaMonkey 1.1.5 Release:
  Most of this week's SeaMonkey work on my side went into the SeaMonkey 1.1.5 release, which went public on Friday, almost in sync with Firefox 2.0.0.8 as planned.
  For my own reference, and for other people to be informed, I documented the SeaMonkey release process along with all this work on the Mozilla wiki.
• Source L10n:
  The Mac breakage is fixed by the patch, so repackaging localized nightlies should actually work. Watch for an announcement in mozilla.dev.l10n soon!
• German L10n:
  I'm still playing catch-up with trunk changes, and fixing bugs reported by testers. Some noticed brokenness in password manager in German SeaMonkey trunk builds, but that is as expected as download manager and windows integration breakage, due to those items still being replaced for SeaMonkey2 and the old ones not being migrated to source L10n.
  Additionally, I released SeaMonkey 1.1.5 German builds in sync with the original US English builds.
• Various Discussions:
  SeaMonkey 2 goals, auto-update system, chatzilla L10n, window icons, feed reading, etc.

There's a lot of exciting stuff going on, and those who don't know it yet might also want to check the SeaMonkey Trunk Tracker blog by Jens Hatlak, which tells about all kinds of interesting stuff happening on our way to SeaMonkey 2 (and maybe even beyond)!

Our team has just released SeaMonkey 1.1.5, the next security update for the stable 1.1.x series that is built upon Gecko 1.8. With this release, we match the updates made for Firefox 2.0.0.8 to also provide suite users with a maximum of stability and security. Get it while it's hot at www.seamonkey-project.org!

As a side note, along with this release, I have documented the current SeaMonkey release process so that others can see how we're doing this, and so that I have written-down guidelines to follow the next time instead of pulling everything out of my fingers every time.

Last week I realized that there's a world outside this room I'm working from and so I spent quite some time with personal activities, so I cut down my working time roughly what normal workers spend at their jobs - still, here's a (better late then never) summary of my SeaMonkey/Mozilla-related work in week 41/2007 (October 8 - 14):

• Branding Cleanup:
  The task of removing Navigator branding from the UI is now done after I got review for the help part and checked that in.
• SeaMonkey 1.1.5 Release:
  We're planning another regular security update to the SeaMonkey 1.1 series with 1.1.5, which is planned to be released at October 18 in parallel with Firefox 2.0.0.8. Candidate builds are up on the FTP server, if QA doesn't turn up anything out of the ordinary, those builds are identical to what we will release. Please help us testing the builds!
• Source L10n:
  As we don't have special machines available for that, I tried to set up our main SeaMonkey tinderboxen for getting localized nightlies through repackaging.
  This worked nicely for Windows and Linux, but the Mac box was broken due to our repackaging Makefile scripts not being able to run correctly in a universal build configuration yet.
• German L10n:
  Another set of Core L10n updates went in, as well as some suite L10n changes German testers found in the first de-nightlies generated by above-mentioned changes. One such problem could be tracked down to be a Core issue, I'm waiting for reviews on its solution. I also could update German READMEs, thanks to Michael for his continued work on those as well as our help documents! Additionally, I made German being included in DOM Inspector builds again after correctly syncing that part of our L10n with the original files again.
  We're keeping German trunk in pretty good shape, even though Filing bugs and getting reviews on every change takes some time, the improvements made due to reviews are usually worth the time spent on them.
• Various Discussions:
  Auto-update system, prefwindow rework, login manager, mailnews development, etc.

As a side note, the Sea-Monkeys in my "think tank" are developing fine, apparently this room is a good place for such animals...

I didn't blog much this week, but things are still progressing nicely. Here's a summary of my SeaMonkey/Mozilla-related work in week 40/2007 (October 1 - 7):

• Window Icons:
  Some discussion and improvements on the proposed new window icons have been done, while some might not like the overall direction I think most people appreciate the long overdue refreshing of those icons.
• Branding Cleanup:
  I worked out the rest of the needed changes for removing Navigator branding from the UI and could also get in the editor part, with the help changes waiting for review and checkin.
• Extension Nightlies (venkman):
  The new venkman team wanted to get nightly builds available, so I did set up a box that reports to the MozillaTest tinderbox page currently, and went through some interesting challenges as tinderbox wasn't originally made for building an extension only.
  I found though that it's not that hard to get up with some slight adaptions to tinderbox code. I also found we had some specialized code for uploading specifically the lightning and xforms extensions when built and created some generic code to handle this instead, which also can be used for venkman in my case or any other extension tinderboxes. I filed a bug and patch for those improvements I'm running, I hope we can get this into official tinderbox source.
  Now the last step left is getting a place to upload to, I hope this can be done soon. We'll then have the venkman that gets built by the Makefile build system uploaded, not the makexpi.sh-generated one, so we might need to make the former one include install.js to provide the same experience, but everything else should work nicely, I think.
  If the ChatZilla team wants, we can also do that for them once it's built as a real extension.
• Linux Tinderbox Update:
  The Mozilla build and IT teams have done some work to upgrade our Linux tinderbox to the new reference platform, which magically fixed a huge startup speed regression we saw when switching from xpfe to toolkit, maybe connected with the newer system tools used by CentOS 5 compared to the older version we were running previously. This may increase the software requirements of our trunk nightly (and probably release) builds, but we're just moving to the same level as Firefox3/Minefield trunk builds, so I hope that's ok for in turn still having nightlies even after checkin of some changes that wouldn't build on the older system (because of e.g. too old pango).
  My job onthis was just to sit tight and watch, and make sure I still have access to the new machine after the change
• Source L10n:
  Not much happened there, but the Linux tinderbox change was something I was waiting for before trying to get our nightly tinderboxen to also do locale repackaging.
• German L10n:
  Some Updates to Core L10n were necessary this week, another one is still waiting for review, and I also needed to do a bit of work to keep suite L10n complete - but we're keeping de pretty well in shape for a diverse team and trunk work
• SeaMonkey 2 Core Tracking Bugs:
  I created two bugs for tracking things in Core we still need or would like to see for SeaMonkey 2. While we can track things nicely with blocking flags in our own Bugzilla components or use whiteboard comments on them, we don't have flags and don't want to pollute whiteboards for other products. As "Core", i.e. all the Mozilla infrastructure we share with Firefox, is more and more in a freezing phase for the 1.9 cycle, we need some way to show what changes there are still important for us to happen.
  The Core bugs causing problems for SeaMonkey 2 tracker points to the heavy issues that we absolutely need or which require nasty hacking on our side to get SeaMonkey 2 out at all (i.e. those bugs we probably would consider blocking-seamonkey2+) - no need to get things on this list though that are already Firefox 3 blockers, though.
  Core fixes that would help SeaMonkey 2 development are our list of things we'd really like to see and which possibly can still make the cut (no place for utopic wishes here), but which are not really necessary for us to deliver SeaMonkey 2.
  If you want to get bugs on those lists, please nominate them with a sufficient description of why they are relevant there. Please only add bugs yourself if you are in the core developmer group of SeaMonkey, and even then add good, but short explanations of why they fit on that list.
• Various Discussions:
  "Turbo mode" removal on trunk, auto-update system, profile switching, prefwindow rework, login manager, Classic and new default theme, mailnews development, etc.

I just started working on the first candidate builds for 1.1.5, so everyone who can help testing, please tune into mozilla.dev.apps.seamonkey and/or #seamonkey in the next few days!

Another week of hard work has passed, here's a summary of things I did in the SeaMonkey/Mozilla area in week 39/2007 (September 24 - 30):

• Window Icons:
  As I posted here earlier, work has been started on replacing the old Mozilla window icons that were derived from Netscape Communicator with a new set that will make SeaMonkey fit in better with your taskbar, window title bars, desktop and other such areas in current windowing systems.
• Branding Cleanup:
  In the icon thread in the newsgroup, someone mentined that the SeaMonkey browser is still called "Navigator" in many areas, and because of that I opened a bug for removing Navigator branding from the UI and already could check in the first part of it.
• README Update:
  When I saw the Navigator branding in the README files we have, i saw that many areas of those are outdated and did a big overhaul of README files following that.
• Profile Location Change:
  Following the wishes of many contributors and discussion with MoFo legal people, we switched our profile locations back from mozilla.org to Mozilla.
• Source L10n:
  Testing the first SeaMonkey builds created from CVS-based localization, our German testers found the editor L10n missing from those builds, which I could now fix as well, thanks for pointing to that problem!
  I also got some nice changes in for making tinderbox locale repackaging more flexible so that we can make our main SeaMonkey tinderboxen to generate localized builds in the future.
  Additionally, I did some testing of dependent language packs for extensions, and found that everything works as expected. I also did an improved second WIP patch for CVS-based ChatZilla language packs following all that testing.
• German L10n:
  I could keep the German localization for SeaMonkey mostly working all the time and could get the first people to test the repackaged builds - once again, thanks for helping with that!
• MoFo ED Search:
  The Search Committee continued the first round of interviews for a new Mozilla Foundation Executive Director. We have an interesting range of personalities for candidates that will be presented to the community in later stages before the final decision.
• Various Discussions:
  "Turbo mode" removal on trunk, community tinderboxen, profile switching, 1.1.x problems with read-only installs, login manager, shell service, stage reorg, UA string flamewar (again), etc.

Things are surely moving in the right direction for SeaMonkey 2. I think we probably will start figuring out soon what we need to do to be able to go Alpha, and what work we will need to complete for Beta and Final.