The roads I take...

Sorry for the delay, got too much to do currently, but here's a summary of SeaMonkey/Mozilla-related tasks I worked on in week 04/2008 (January 21 - 27, 2008):

• Automated Update Service:
  We tested SeaMonkey AUS support and it looks to be at a state where it works well enough that we can use it to update users from a first alpha to a later (alpha/beta/final) build, so the part that blocked alpha is fixed. Further improvements might be nice, but not strictly required at this point.
• Lightning supprt:
  With my recent changes for making Lightning work with SeaMonkey, it seems to get somewhat usable. Further bugs about things that don't work should be filed in Bugzilla's "Lightning: SeaMonkey Integration" component.
  BTW, we know that missing toolbar customization in MailNews is the cause of a few of the problems we still have with this integration, but that's not too easy to fix right now - let's see first how well that moves forward for the browser component.
• German L10n:
  Nothing more here than the usual set of updates for trunk.
• www.m.o website:
  I took part in a www.mozilla.org planning meeting, where we discussed how to make the main Mozilla project website fit for the future - or at least match reality again. The target is making it a community portal as well as the place for official Mozilla project content. See the linked post for more discussion.
• Various Discussions:
  Freezes, kill-mork and kill-rdf in MailNews, FOSDEM, etc.

I'm spending lots of time in IRC discussions these days, about ongoing MailNews backend work and different core and SeaMonkey development topics, most of which doesn't result in very "productive" things to list here, though.
And one more thing: We'll be trying to get up candidates for a SeaMonkey 1.1.8 release this week and hope it will be tested well - due to publication of an exploit for a security flaw already fixed for this release, we'll be in a tight schedule to release this one.

Here's a summary of SeaMonkey/Mozilla-related tasks I worked on in week 03/2008 (January 14 - 20, 2008):

• Tinderbox and Extensions:
  I checked in my patch for being improvements to tinderbox extension handling, Mozilla boxen compiling Lightning and XForms are using that code instead of hacks specific to those extensions now. My tinderboxes that generate venkman and ChatZilla nightlies also use that code now, unfortunately we still miss a directory to upload such extensions.
• Automated Update Service:
  SeaMonkey AUS support should be working with nightlies for the most part now, as I turned on pushing of update info to the server and Mark added UI support for updates later in the week.
  This means that now you should be able to check for updates in nightlies, as well as download and install full updates. We don't create "partial" updates for nightly builds yet, as this requires more more on the nightly build system (the Mozilla build team is working to get that for all our software), and the issue with updates for localized builds on Mac is still unresolved. We also may have some other glitches, but most of the work seems to be done here to get what we need for Alpha in this area.
• Lightning supprt:
  I did take a lot at the errors Lightning is producing when turned on in SeaMonkey, and did find a few small changes we could make to come nearer to making Lightning work with SeaMonkey. There's much more to be done here, and I'm not sure how much I can and will work on this, but when my patches are in, we can at least switch between calendar and mail modes as well as display and edit calendars and events. Note that even this needs trunk code of both Lightning and SeaMonkey, plus the patches I submitted to the bug.
• Source L10n:
  I landed CVS-based venkman language packs, and could already add support for German, French, Russian and Swedish for this JavaScript Debugger extension. Later in the week, I also added support for those language packs to the SeaMonkey Windows installer, which now installs the correct language pack (if it exists) when (and only when) the user installs venkman - which is the default but can be unchecked.
  I also filed a bug for AMO support of such extension language packs.
  In other "source L10n" news, I added Norwegian to the list of SeaMonkey trunk locales, which makes us produce trunk nightlies in up to 10 languages (including en-US) now - as long as localizers keep their translation updated in CVS.
• German L10n:
  I landed German support for venkman, which now makes SeaMonkey trunk builds being completely localized (except download and password managers, which just don't work in localized builds currently). Additionally, some other work was needed to keep trunk up and running, which I also too care of. Finally, Micheal Opitz contributed another Update of German SeaMonkey help, thanks for that.
• Mozilla Foundation Executive Director Search:
  We interviewed three more candidates this week, it becomes clear over time that this process takes longer than expected, as it's already quite hard to find the 3-4 people we want to take to a second round of interviews. People who are leader types, understand open-source and community processes very well and are willing to take this job are obviously rare.
• Various Discussions:
  Preferences window migration, ChatZilla as optional installer component, feed support, www.mozilla.org website, FOSDEM, etc.

There are two thing I want to add at this place:
I have turned off nightly build cycles for the Mozilla 1.8.0 branch on my private machines, so no SeaMonkey 1.0.x nightlies are being built any more. We haven't intended to do any more release of that series for some time now, stopping to produce nightlies about 7 months after the last release from that branch seems to be a lot of time for people to adapt to not using those builds any more.
The second item I want to shed some light on is the Mozilla Meeting at FOSDEM 2008 which will take place on February 23/24 in Brussels, Belgium. I'm sure this will once again be a good place to meet many of our fellow Mozillians - and for those who want to hear about things going on in community project management and SeaMonkey 2, I will give a talk titled "And the beast shall come forth..." - Hope to meet you there as well!

I've just read an interesting blog post about rewrites in MailNews, which would improve both Thunderbird and SeaMonkey by replacing very long-lived, historically grown, patched and hacked interfaces in the mail and news code by interfaces that cleanly apply to present architecture and are again open for relatively clean additions in the future. This doesn't mean the current code was that bad at the time it was created, it was made for the state of mailnews in those times, and those times being almost 10 years ago (early Mozilla code) or even before (code we inherited from Netscape) clearly explains why things like message tags or feed account types require some sort of hacks to apply cleanly to the old code. Also message meta info being stored at folder (newsgroup) level instead of account level means that it's e.g. hard to mark cross-posted message read in multiple newsgroups. A side effect of those reworks is also that old storage formats like Mork databases and RDF datasources can be replaced by more modern backends like the SQLite-backed mozStorage mechanism.

Joshua Cranmer, who wrote the blog post I cited above, is currently trying to get a picture of where we want to end up with such rewrites, how we can get rid of Mork ("kill-mork" work) and many RDF usages ("kill-RDF" work), and how to design interfaces in a way that they are fit for future enhancements.
When the picture gets clearer, he intends to work on kill-mork tasks in address book first, from what I read.

I am looking forward to seeing this work evolve, I think it could improve some of the code areas that many people are unwilling to touch because of their historically grown clumsiness - and in open source, it's always bad if people refrain to touch code because they don't fully understand it.

Here's a summary of SeaMonkey/Mozilla-related tasks I worked on in week 02/2008 (January 7 - 13, 2008):

• Tinderbox and Extensions:
  Philipp Kewisch came to work on the same subject as I had some time ago, being improvements to handling extension updates in tinderbox and removing the specific hacks we had in place for Lightning and XForms.
  Well, he actually looked into Lightning, I had looked into running a tinderbox for venkman, maybe even ChatZilla. His first step landed with slight modifications, I'll doing a second step in my "old" bug.
• Automated Update Service:
  I filed a patch to get us complete updates for Mac L10n builds. We'll try to make progress on the actual SeaMonkey AUS work this upcoming week.
• Artwork:
  As we support using PNGs for GTK window icons now, I worked on converting SeaMonkey Linux builds to doing just that.
• Extension Interfaces:
  You might know the FUEL JavaScript library designed for Firefox extension developers, maybe also the plans on doing something similar for Thunderbird under the name of STEEL, for both of which Joey Minta has done some cool work.
  When discussion on IRC once again came to that topic, I proposed something like this should also be created for SeaMonkey, with which Joey agreed, and thinking about it a bit more, I proposed it could be named "SeaMonkey Interface Library for Extensions" or SMILE and filed a bug for that. Our team currently doesn't have much time to work on such a library but tracking this is a good idea, I think - and we'd appreciate help on that topic.
• Source L10n:
  The CVS-based venkman language pack patch is waiting on cvs copies, I should have poked the right people so we can get those done.
  I also could add Belarusian to the languages we support on SeaMonkey trunk and Swedish to the supported ChatZilla locales.
• German L10n:
  I kept German SeaMonkey trunk green with the usual amount of small updates to SeaMonkey and Core L10n files.
• Various Discussions:
  Preferences window migration, Vista support, SeaMonkey Trunk Tracker on Planet Mozilla, Sidebar lists, www.mozilla.org website, language switching in EM, etc.

Development on SeaMonkey 2 has been picked up well again this year, and most of the Alpha blockers are seeing some progress. I may start providing irregular updates on status of our work towards a first SeaMonkey 2 Alpha release in the next weeks.

The first week of the new year still was dominated by holiday and New Year celebrations, but I managed to get some SeaMonkey/Mozilla-related work items completed which had been waiting for some time, here's a short summary of what I could do in that area in week 01/2008 (December 31, 2007 - January 6, 2008):

• SeaMonkey Project Organization:
  Finally could post to the official SeaMonkey blog about the SeaMonkey Council Restructuring (CTho left, but ajschult, Mnyromyr and Standard8 have joined us instead).
  I also finally posted the SeaMonkey 2 Alpha Criteria here on my blog, and marked the most grave showstoppers as blocking-seamonkey2.0a1+.
  Both those events should help us getting a good start into this new year and pave the way for getting a good major release in 2008.
• Nightly Build Optimizations:
  I switched the SeaMonkey tinderboxen to use the default optimizations instead of explicitely setting them in the mozconfigs, which enables us to use different optimization settings for different parts of the tree and balance binary size vs. performance much better. Firefox is doing the same now, and improvements of the in-tree default settings are under investigation.
• Source L10n:
  I came around to port the CVS-based language pack approach to venkman, so we hopefully should get that extension (optionally) localized from source as well soon. This will enable us to ship fully localized SeaMonkey 2 releases directly from files in the CVS repositories.
  I also found a small problem with update packaging for localized builds on our Mac machine, I hope we can fix this quite easily.
• German L10n:
  A few small updates had to be made to keep SeaMonkey green, but nothing spectacular.
  I locally prepared a current German venkman localization to test the work stated above and will land this when the main patch goes in.
• Various Discussions:
  Preferences window migration, thoughts on future of address book, web discrimination, etc.

I also managed to post a long-prepared statement on web discrimination right at the end of the year, and I hope the Mozilla story of 2008 increasing their focus on spreading what the Mozilla Manifesto is about will also help us to spread the word about this injustice in today's web landscape.

Since our rework of the SeaMonkey suite to using the "new toolkit" on trunk is making good progress, the SeaMonkey Council has talked about setting up criteria our trunk builds have to meet so we can release a first Alpha for SeaMonkey 2.

The relatively small and compact list of those criteria is:

1. AUS (automated update service) needs to work,
2. L10n builds need to be usable,
3. no major broken functionality,
4. profiles created or migrated with Alpha need to be in a state that doesn't need re-migration until final.

We agreed that not having two preference windows (like we do have in current nightlies due to the ongoing transition to the new one) is along the lines of c) here, and b) needs password and download manager work, maybe even some shell service stuff.

Being feature-complete or doing any kind of feature or locale freeze is something for the Beta phase, rather than for Alpha.

Point a) in this list mainly means that users of Alphas should be able to utilize AUS for upgrading to later builds, esp. Beta and final builds.

The main requirement for d) is that users who migrated a profile with Alpha don't need further "manual" migration to keep using this profile (or even throw away and re-migrate it) with later builds in the 2.0 series, including final and minor updates.
Things like changing to places history should automigrate existing data, so there's no problem with doing that post-Alpha, a small question may remain open for how to deal with people migrating from Thunderbird and having feeds in those profiles, as we probably want to have them pick those up once we support them, but they may look strange in mailnews currently.

The SeaMonkey Council will start marking Alpha blockers according to that criteria in the next days. Note that minor annoyances probably will not block an Alpha release, though we'd like to see them fixed. We're not targeting a fully cleaned up final release yet, this is still an Alpha and it's neither supposed to be feature-complete nor regression-free, but ready to be tested by a significantly wider range of community members than our usual trunk nightlies.