The roads I take...

Here's a summary of SeaMonkey/Mozilla-related work I've done in week 03/2010 (January 18 - 24, 2010):

• Build Infrastructure:
  A lot of more work on L10n nightly updates until I finally got then to work. Note that any SeaMonkey nightly updates are complete updates only, as the tooling to do partials for nightlies on our normal build machine pools is not done yet, currently waiting on Mozilla RelEng there. Getting our nightly updates to work correctly needed changes across 3-4 repositories (including mozilla-central and comm-central), but everyone was cooperative and it ended up great.
  The next step I had on my mind was switching to packaged tests, both for being nearer to what Firefox run and making it possible to run test suite in parallel, esp. as mochitest-plain runs really long. To my surprise, this worked well, I needed only one small change to the generic test factory and it worked for SeaMonkey (I later coded up an improvement to report Mozilla revisions, but things worked fine without it) and tests did run fine except for a mailnews xpcshell problem, which I hope will be dealt with soon. With that, I could officially do the switch for the most part.
  I also did experimentally create some hourly builders to run on a 1.9.2 tree, which we don't really want to target for SeaMonkey, but want to at least build fine. I spotted a packaging problem that was a regression form a recent build system port and Serge swiftly fixed that one - thanks for that!
  And, as I now know how to deal with packaged tests and in the new architecture we should run them on normal debug and even optimized builds, I turned on the xpcshell test suite for SeaMonkey trunk Linux builds (the debug ones we do for leak testing), even if a number of failures show up there - it's the first step to fixing them. A one-cycle test run of the other suites pointed to some more issues to fix, but we don't have the machine power right now to run them all the time.
• Download Progress Windows:
  After some proposals have come up for fixing progress window shortcomings, we were looking for icons that were usable under our licenses, and when those failed to materialize, I went and created my own SVG icons to match the small ones we have right now, so I could produce larger versions. Having done that, I worked on patching the dialog and posted screen shots of this work in progress in the bug.
  Back when I did the version of those windows that is in 2.0 right now, I did the fastest solution I could while still applying a design facelift, now for 2.1 we should have the time to improve on that and correct problems we see with this 2.0 design.
  I also updated my patch for moving core build master code to a shared place.
• History Expiration:
  I'm risking the next flame war against me with this, but I did a patch to react to the places expiration rework done for the whole Mozilla platform, which also removes the ability to limit history to a fixed number of days or sites, but instead goes with a memory percentage. We will need to implement the cleaning up private data based on time frames to complement this to at least reduce the complaints, I think. While working on this, I saw that Thunderbird also has prefs for this around and filed a bug for them as well.
• SeaMonkey L10n:
  As mentioned above, L10n nightly updates for SeaMonkey trunk work now, I posted to the L10n newsgroups about this as well.
• Various Discussions:
  Add-on compatibility communication, new machines, comm-central policy for requiring tests, nightly.m.o, possible platform roadmap options, Firebug 1.5 release, Firefox 3.6 release, external linkage for mailnews, KompoZer and SeaMonkey, YouTube and "HTML5 video" vs. Ogg, community-based geolocation service, etc.

This has been a really productive week again and it feels good to get real things moving and also start to do work and planning for SeaMonkey 2.1 now, turning the head back and putting out fires on 2.0 was really getting tedious - even though we have a slightly conservative approach here in SeaMonkey land, we are as much about progress and innovation as the rest of the Mozilla project (even if it is in our way and sometimes means the some changes are not as much into-your-face and revolutionary as in other projects but have more of a continuity label on them).
I hope we all can get into this more again now and get some exciting patches landed for the next version of our great suite.

Here's a summary of SeaMonkey/Mozilla-related work I've done in week 02/2010 (January 11 - 17, 2010):

• Releases:
  SeaMonkey 2.0.2 was released on Monday, mainly fixing a Windows mailcompose freeze which was a regression from 2.0.1 work. The previous 2.0.2 scheduled for early February was renamed to 2.0.3 and moved to the middle of that month.
• Build Infrastructure:
  I finally attacked the problem of L10n nightlies not being triggered and for that moved right over to mirroring the Firefox setup as closely as possible. After fixing a glitch from some wrong variable-repurposing for WinCE build work, it was a success in that our new config allows us to go for things like split test cycles much more easily and that L10n nightlies are being produced, but I needed to split L10n nightly updates into a different bug and do some more work on that - and we're not completely there yet.
  Also, the release automation bustage fix found with 2.0.2 work could be checked in, as well as the switch of release automation to pulling chatzilla from hg, which I now found time to write a patch for.
• SeaMonkey L10n:
  I finally came around to activating Italian ChatZilla and venkman, and Ukrainian SeaMonkey.
  As said above, work is going on for L10n nightly updates - right now, everything is actually working except that a wrong patch is being written into the snippets and so no update can be installed, even though builds know that it's available. I'm working on fixes for that (needs work across three Mercurial repositories).
• German L10n:
  To have something for reasonably testing the L10n building updates on trunk, I wanted German to build alright, so I went in and finally updated ChatZilla and venkman localizations, and while I was at it, DOM inspector and SeaMonkey as well. I also did small updates to DOM and security to sync them to current trunk development, but toolkit still needs a bit more work to be green on trunk.
• Various Discussions:
  Add-on compatibility center, new machines, comm-central policy for requiring tests, plans or no plans for platform releases, Lightning 1.0 Beta 1 release, external linkage for mailnews, Manifesto and privacy, profile management future, KompoZer and SeaMonkey, etc.

Work is definitely picking up, and I've also started trying the Mozilla Status Board tool, which will not replace but be an addition to my updates here - with the benefit of having a section of what I'm planning to do next. And my list of "next" items was quite large last week, I'll probably need to keep a few on that list this time as well...

Here's a summary of SeaMonkey/Mozilla-related work I've done in week 01/2010 (January 4 - 10, 2010):

• Releases:
  I prepared a 2.0.2 update this week, containing a very small set of fixes over 2.0.1, most importantly a fix for freezes in composing emails on Windows when the OE Contacts address book is present.
  Along with that short-cycled update, the previously planned 2.0.2 release scheduled for early February was renamed to 2.0.3.
  Unfortunately, the automated release process didn't perform completely without problems this time, I ran into a timeout with our mini as well as some bustage from a recent infrastructure update and did a patch for the latter.
  In the end, I got candidate builds and beta channel updates out on Tuesday evening, and could prepare all website changes by Sunday for a Monday release.
• Add-Ons:
  We're trying to encourage add-on authors to make their work compatible with SeaMonkey 2.0. Most of the work for that is being done on our side by Philip Chee, e.g. for sending out messages.
  We also would like to get an add-ons compatibility center set up on AMO though, and I prepared "SeaMonkey 2.0" wordmark images for that.
• SeaMonkey L10n:
  After the 2.0 release, we have been pointed to a license problem with packaging dictionaries into language packs, which we worked around in the recent releases, but for the future (beginning with 2.0.3), we're not shipping dictionaries in langpacks to resolve this. We now are only packaging the chrome localization, perfectly matching what Firefox does. This also makes it easier to merge more of the build processes across Mozilla applications in the future.
  If you are using language packs and want the matching dictionary installed in the future, fetch it from AMO - which makes it even available if you switch between languages.
  The fully localized builds are not affected and still contain the matching dictionary (if available under the MPL from Mozilla repositories), by the way.
• Various Discussions:
  L10n build infrastructure, close button on tabs, build system porting, AIX port, extension dependencies, future of theme and extension systems, external linkage for mailnews, etc.

With my dad's birthday in the middle of the week, I once again spent most of the week at home with my parents and didn't get around to a whole lot of work, but still tried to start picking up the pace at least somewhat - with a priority on shipping the mail compose freeze fix to Windows users in the 2.0.2 update.
As of right now, I'm back to a normal working schedule, and starting to think more and more about where to go with 2.1, both in terms of features and code work as well as timing. Unfortunately, I don't see the Firefox "Lorentz" story finalized as of now, but timing of the next Gecko/platform releases plays a critical role in our own release planning. I hope to see some light shed on those matters soon.

Here's a summary of SeaMonkey/Mozilla-related work I've done in week 53/2009 (December 28, 2009 - January 3, 2010):

• Build Machines:
  While cleaning up some part of my email structure for the switch to the new year, I finally filed a few bugs on things I've had on my mind for some time now:
  Not shipping dictionaries with langpacks was the outcome of a license debate after the 2.0 release, I also did a patch for that but didn't come around to checking it in yet due to being on and off all the time currently.
  Once we know we won't ship releases from a point before the switch of ChatZilla to hg any more, we can also switch releases for that, but 2.0.2 has been short-cycled based on the pre-switch 2.0.1 release.
  Some localizers pointed to L10n builds not being generated daily, I need to figure this out soon, possibly together with L10n nightly updates.
  Since the recent Parallels upgrade, one of our Mac VMs has been running in the original 2-CPU configuration again and didn't show problems, so we can switch back the remaining ones as well. If that turns out fine, the Parallels experiment might still find a positive outcome.
  That could very much influence getting machines for a third SeaMonkey tree next to trunk and 1.9.1, as Parallels might be a viable option again now.
  And I requested signing infrastructure though I'm not yet completely sure what exactly we need there.
• German L10n:
  Just in time for the 10-year anniversary of German Mozilla releases, I dug out the first packages I created for M12 and M13 as well as the (English-only) announcements and put them up on the German SeaMonkey website so other people might indulge in nostalgia as well.
• Various Discussions:
  Windows 2.0.1 freeze when entering addresses (related to OE contacts), profile manager UI, external linkage for mailnews, Callek coming back to the intarwebs, etc.

This was another week in which I kept a low profile on work, spending some time with friends and doing some maintenance on my data - all of which are things I hope help turn out things well in this just-started year, which will host a nice anniversary for the SeaMonkey project - we're turning five years old!