The roads I take...

Here's a summary of SeaMonkey/Mozilla-related work I've done in week 16/2010 (April 19 - 25, 2010):

• Automated tests:
  Looking through my local patch queue, I decided to pick up one again that had been lingering around for quite some time - the context menu test, including its checks for duplicate accesskeys.
  I updated the patch and got review fast, so I checked it in. As I had somewhat suspected, it produces failures in Windows due to "Set As Wallpaper" on images only being present there right now. When I worked on fixing this, the whole can of worms that those accesskeys present opened again, and Neil started to propose removing some entries to fit accesskeys better. We finally settled on taking my current fix as I have it and take a look at what entries are reasonable in what cases in a followup.
  I also discovered that entries where we automatically change labels can end up empty in case of coding errors, so I added a check for that in the test, and ported that check to the Firefox version as well.
  I also discussed a few of our remaining test failures with Serge, I hope we can get all those out of the picture soon.
• Internal Help Content:
  Help updates for places history have now landed on 1.9.1 branch as well, so SeaMonkey 2.0.5 will have that. The (trunk-only) expiration-related help update was reverted to a positive review and so I checked it in.
• Using Services.jsm:
  After I saw a patch for using it in Firefox nsBrowserGlue landing, I decided to file a bug and create a patch for using Services.jsm in nsSuiteGlue. This is a nice cleanup and code readability improvement, got fast review and I checked it in. I also updated the nsSuiteGlue parts of my places work for this locally.
• Project Organization:
  I did some more work on project area tools, now we can dynamically manage a copy of the project areas list on my SeaMonkey development site. This has roughly the same data as the official list right now, but is a work in process and only meant for administration, not as the official reference.
  We discussed SeaMonkey 2.1 Alpha 1 planning in this week's status meeting and settled for May the 4th as code freeze for a first alpha. We tentatively put a second alpha and a single beta up as planned milestones in 6-8 week cycles after this first step. A final release can not be scheduled without knowing more about Mozilla platform 1.9.3 plans, though.
  For managing bugs for and planning those release, I have files a bug on new-style flags for SeaMonkey 2.1 but I underestimated how much work there is behind creating them, so they are not in place yet (should be near, though).
• Bug Triage:
  The Mozilla bugday on SeaMonkey "General" bugs on Tuesday was a shining success, I did help with a few bugs on the list, but the awesome people there did an amazing job and greatly reduced the amount of open bugs in the general component! Thanks a lot, guys!
  In the status meeting, we also decided that we'll go forward with my bug expiration proposal - with the exception of help bugs - and I posted a comment text I'd like to use but got no feedback on that yet.
• Places:
  I did some small local updates to my places bookmarks patchset, as well as ported some changes from the Firefox side to it, while feedback from the new try builds is good, some small issues reported are double-click on bookmarks in the manager not working, HTML import from bookmarks manager being broken, and personal bookmarks toolbar buttons wrongly reacting to "icons only" customization settings.
• German L10n:
  I landed a relatively large help update from Michael Opitz that brings 2.0.5 and trunk contents up to sync with what we have in the US English version.
• Various Discussions:
  Tabbrowser APIs, review speeds, Mnenhy, "site" vs. "web site" vs. "website", SeaMonkey 2.0 interview series, master password hack, comm-1.9.2 branching, drumbeat catchphrases, hackers and posters, etc.

So, the cut for the first alpha of SeaMonkey 2.1 is nearing, but we're still missing some flags for managing it and my rather large patchset that would do good with testing there will very probably not get reviews fast enough.
Still, a lot of good work is happening, people are putting hard work into our code and we have a ton of platform changes to show off. I think we're on a good way, we just need to pick up a bit more work in our own code, and we're of course happy about any help on that - as well as in areas like marketing, QA, and design!

The releases wiki page is still stating May 4th as release date for Firefox and SeaMonkey security updates, though those may shift a bit. Also, Thunderbird 3.1 Beta 2 is planned to go out that day - and our next SeaMonkey Status Meeting should also happen that day. But all that is not what this post is about.

In today's SeaMonkey Status Meeting, we discussed if and when we should freeze for and ship SeaMonkey 2.1 Alpha 1 to show people things are moving in our project and to get code out to testers. We collected what changes we already have in trunk since the SeaMonkey 2.0 were cut, and the Mozilla platform has given us a ton of new things alone, ranging from speed improvements (in startup, Js, etc.) via privacy improvements (good-bye :visited sniffing!) to new features on the web content (WOFF, SMIL, and more) and platform/development side (js-ctypes, pymake, etc.) - and then there's mailnews work and already some SeaMonkey-specific improvements as well. In short, a lot of things to show to our testers!

Because of that, we decided to put an important date on our schedules:

SeaMonkey 2.1 Alpha 1 code freeze: May 4th, 2010, 23:59 PDT

We'll start builds as soon as possible after that, reopen the tree for 2.1a2pre development and release 2.1 Alpha 1 once we have enough testing (tentatively May 18th, possibly earlier).

So, if you have any code you want to get out to testers in SeaMonkey 2.1 Alpha 1, make sure it lands within the next two weeks!

(And thanks to InvisibleSmiley for the tagline of this post!)

Here's a summary of SeaMonkey/Mozilla-related work I've done in week 15/2010 (April 12 - 18, 2010):

• Build Infrastructure:
  I had some discussions on keeping build machines up to date and getting a clobberer.
  As those things came up in my thought or discussions, I filed bugs to track potentially testing on opt builds and having some automated perf testing. Nothing is planned for the immediate future on either, but they should be tracked so we see needs and possibilities somewhat better.
• Bug Triage:
  Following my expiration proposal, discussions continued, results are not conclusive yet.
  And don't forget, there will be a Mozilla bugday on SeaMonkey "General" bugs tomorrow!
  I for myself took a peek on the components table I linked in the possible triaging targets post and did a pass through all open installer bugs, resolving a lot of them as INCOMPLETE as the xpfe installers are dead after the SeaMonkey 1.x EOL.
• Internal Help Content:
  My patch that updates help for places history got reviews and landed on trunk, it's awaiting branch approval now. On the expiration-realted update, I'm still a bit unclear about the r- comment I got.
• Project Organization:
  We currently have no good component for history bugs, I filed a bug for finding a home for them.
  Google Summer of Code is in the stage of assigning slots to projects and choosing which projects to go with - the good news is that we probably have a project on Weave and SeaMonkey, I'm mentoring that one.
  I did a lot of work on tools for bringing project areas up to date, e.g. analysis on patch authors, reviewers and bug components for hg history of all suite/ files, see the project files list on my SeaMonkey development site for that data.
  On the development group/list, I posted about SeaMonkey 2.1 Alpha 1 planning, but unfortunately didn't get much input there yet.
• Places:
  I updated all my places patches this week, but I decided I'll not update history for anything else then the existing performance work and breakages until bookmarks land and history is switched to use shared core files with that new work.
  The places bookmarks patchset is now up to everything that has landed on Firefox, and I have new try builds up for your testing, preliminary reports tell that they are working as well as I hoped, the only bug I know of is double-click on bookmarks in the manager not working, but context menus allow to open them in new tabs or windows just fine for now.
• German L10n:
  I checked in a help patch that was long overdue for landing.
• Various Discussions:
  Test failures, tabbrowser APIs, about:rights and toolkit, browsehappy.com, new committers and sr requirements, review speeds, Mnenhy, etc.

A lot is going on, and as there have been 4 developer previews of the Mozilla 1.9.3 platform already, which will form the base of SeaMonkey 2.1, I think it's time to think about actually releasing a preview of our own and get a first alpha going. My feeling is that we have a few interesting things in current trunk code already, but we need to collect some data to be sure, please reply to my post on the development group/list if you know things important to note in what current trunk has in comparison to the stable 2.0 series.

Oh, and be sure to show up on the bug day if you can help us in any way there!

I did it again - this time, exactly one month after the initial posting of a full patch set, I updated the places bookmarks patches, now even including the latest work that happened on the Firefox side.

And here are the round 3 try builds from those new patches:

• Linux tar.bz2
• Mac OS X .dmg
• Windows .zip
• Windows installer .exe

As before, please test esp. bookmarks handling in those builds, ideally with copies of real-world profiles (they might destroy anything as they are highly experimental, don't use them with any profile that has no backups) and report any issues you're seeing.

Here's a summary of SeaMonkey/Mozilla-related work I've done in week 14/2010 (April 5 - 11, 2010):

• Build Infrastructure:
  Partial updates for nightlies could be actually made working after fixing the regression I introduced on first try. For any 1-day update jump on trunk nightlies of any locale, you should now get partials via the automated update service, only larger jumps need to go through complete updates now.
  In all that work, I also updated the community update server to current AUS code so that our matches all the features of the "official" one again.
  Once again, I also upgraded our buildbot master to current buildbotcustom code to make sure everything continues to work smoothly there.
  Not enough with that, I also turned on test packaging for optimized "dep" builds on trunk so it's easier to download and run the tests yourself, even without building on your own machine.
  Once again, I also did some build system reviews, and discussed a few test failures in bugs.
  As a fallout from the update discussions, I filed a bug on updating our AUS URL. And before I got to ranting more that it's not easy nowadays to keep our build machine configs on par with the Firefox ones, I filed a bug for making it easier again.
• Bug Triage:
  I already posted last week on possible triaging targets, but this week I did take another look of what we could do to deal with bugs more easily and in the end proposed expiring bugs that had been mass-unconfirmed 10 months ago and had no comments since then. That step would make it easier for our small community to concentrate on bugs that had some activity since our project has been started.
  There also will be a Mozilla bugday on SeaMonkey "General" bugs, which should be checked and probably moved to a more appropriate component.
• Internal Help Content:
  As the whole interface change had been introduced by me, I volunteered on this week's SeaMonkey meeting to update help for places history and did that work later in the week, wondering how little work it actually required.
  While I was on the topic, I realized I also should tend to a related trunk-only help change.
• Places:
  Some side discussion came up when it was discovered that some related part that wasn't really needed there but which I included in my making icons work WIP patch was actually useful elsewhere - and possibly rather now than post-places-bookmarks. I urged people to decouple issues as far as possible, and if we really need this now, implement it in a separate bug.
  On the main places bookmarks patchset, there's not much to tell, other than the recent try builds are being reported to work nicely. More testing wanted!
  Meanwhile, I filed bugs on getting some tests in place, updating help and UI prefs updates, which all can be done after the core of the work itself.
• Various Discussions:
  Build system reviews, test failures, bugday, Google Summer of Code, tabbrowser APIs, Firefox tabs-on-top experiment, SeaMonkey 2.1 progress, etc.

I'm continuing and intensifying my thinking about how we can get more people to help the SeaMonkey project. We desperately need help in marketing, support, design and bug triage areas - which all are not tied to very in-deep code knowledge but even easy to work on without any programming knowledge. If you can help us out there, please contact us - and if you know someone who can help us, please spread the word!

Here's a summary of SeaMonkey/Mozilla-related work I've done in week 13/2010 (March 29 - April 4, 2010):

• Releases:
  Shipped SeaMonkey 2.0.4, including work on finalizing release notes and getting announcements up and out the door.
• Build Infrastructure:
  Worked on making partial updates for nightlies available and got them almost to work, if I wouldn't have introduced a general regression. This also needed one patch for buildbotcustom to slightly fix up things from the original work.
  I also did some updates to build-configs/seamonkey to make it even simpler and more similar to mozilla2.
  After that, I made SeaMonkey run mochitest-plain in chunks on trunk.
  I also did some more build system reviews, and extensively discussed some test failures.
• Organization, Website:
  I had some discussions this week on geolocation support, both with the GLS contact at Google and the Mozilla contact - unfortunately, without any immediate outcome. We might want to consider finding an alternative to get our geolocation info from.
  When I saw that not too much is happening in the area of bug triage, I posted about where and how we should pick up triaging again. On April 20, Mozilla will also hold a bugday on SeaMonkey, we still need to decide the specific topic for that, though.
  Now that we have components for release engineering and the website, I triaged bugs into them and tried to get some more resolved.
  And as I already was on website stuff, I updated the SeaMonkey website to feature a JSON with current versions, reduce places to update when releasing a new version, and a page with links for downloading nightlies.
• Places:
  I sat down and did the bare work to making icons work with places bookmarks, even if this is just a work-in-progress patch, it's good enough to use for the moment to see how things can/will work.
  Thanks to some tests from others and my own test work, I found a few slight glitches in my main places bookmarks work (but ones with partially large effect), so I created another round of try builds with those fixes and the icon patch. Those should be really nice for testing now - but be aware that they are worse than trunk nightlies, they are from some point on trunk as well, but with an experimental patch set on top.
  I still need to find a solution with Neil how to get this work into the tree without taking overly long for that.
• Misc Code Work:
  The patch for reworking the branding directory had been out for some time, I did one more update to it, got a review and landed it.
  Digging into automated tests, I found that we were using undefined URIs that might even have come from my wrong saying in former times, and did a fix to replace resource://app/ with resource:/// that also already landed right this week.
  I also created a workaround patch for making "about:" actually useful.
• Various Discussions:
  Build system reviews, test failures, one more 1.9.2 chemspill, security group structure/access, MPL update discussions, reducing my TODO list, getting 2.0 contributor interviews going again, lift10, etc.

One large new thing for myself is that I finally made the jump in my personal builds from US English SeaMonkey 2.0 builds to German trunk builds - including all my patches, that means places bookmarks among other things! For one thing, that showed me that importing the old bookmarks into the places system works fine and I hardly see that I have a different bookmarks system when doing daily work, as it works just as nice as it can.
I'm using L10n-merge on the local machine to augment the German strings with stuffing in US English ones in the holes, but that way I'm now able to eat more dogfood, which should help all my trunk-oriented work. Oh, and I found some nice trunk features as well, including this nice resizer that let me make this text area in my blog large so I can write larger tests more easily!

I have posted about places bookmarks try builds before, and meanwhile found a few flaws that broke a few things in the first round, have a WIP patch for getting icons set in the database when pages are visited, and ported some small fixes made to the Firefox side of places.

I have just moved to using builds with places bookmarks for my daily browser usage (and ones with L10n-merged German L10n, which should make dogfooding even more fun), so I hope I'll find out about some of the possible issues myself, but more testing from the community is highly wanted!

And here are the round 2 builds:

• Linux tar.bz2
• Mac OS X .dmg
• Windows .zip
• Windows installer .exe

Once again, please test those builds, ideally with copies of real-world profiles (they might destroy anything as they are highly experimental, don't use them with any profile that has no backups) and report any issues you're seeing.