The roads I take...

Here's a summary of SeaMonkey/Mozilla-related work I've done in week 21/2010 (May 24 - 30, 2010):

• Releases:
  Still waiting on new Firefox updates to be so we can do a sync release of SeaMonkey 2.0.5 as well.
• Build Infrastructure:
  Watched the trees for intermittent failures and starred a number of them. We still have 2-3 Windows slave losses per day, but it's much better than before the slave updates for log buffering.
• View Source:
  My patches for both using toolkit's view source consistently the toolkit fixes have been reviewed and landed, we now have consistent and all suite-style view source windows.
• Add-ons:
  I've done a real lot of add-ons-related work this week, starting with bumping the maximum SeaMonkey version to 2.1a1 for Sync and Account Manager.
  While we are on sync, Google Summer of Code has taken off and Harini has posted a first result of her work in the bug to make tab sync work for SeaMonkey. Going through the reviews should make her more acquainted with the code style customs around here and set up the stage for the work on a mailnews metadata engine later this summer.
  The rest of my work in that area this week had been triggered by some recent discussions including creative thoughts on the future but also other topics causing some frustration that I had decided to escape by putting in some work and generating some sense of success. I came to the conclusion of converting some long-going thoughts of mine into concrete action and create a new Data Manager targeted for inclusion into SeaMonkey 2.1, if all goes well.
  While I was to create this as a new add-on first, to enable rapid development, get better testing and make it even available for Firefox, I realized that making my long-standing Mandelbrot XULRunner application available as an add-on would make it easier for people to download and test it, which is still quite clumsy for XULRunner applications.
  That in the end, resulted in adding two new add-ons to addons.mozilla.org, namely KaiRo.at Mandelbrot and Data Manager.
• Places:
  After I updated the Firefox fixup patch, Mano had time to review it and I could check it in. That also meant that the last question on the places history speedup patch was answered, a review could be completed and I checked that in as well.
  With that, the major prerequisite for places bookmarks is right there in the trunk tree! Following that, I created a new set of patches and yet another round of try builds that also should fix all the bugs reported to me so far with the previous ones. I hope we'll be able to get actual reviews of those patches soon.
• SeaMonkey L10n:
  Related to L10n at least is the topic of making "site" vs. "web site" vs. "website" usage consistent, and the "official" word on that seems now to be that "website" is the official choice for anything in that realm that can be said to be "on the web". Because of that, I created a patch to convert almost all of SeaMonkey to use "website" where "web site" or "site" had been used to talk about websites.
• German L10n:
  While thinking about how we could get more knowledge of what Mozilla work is happening in German-speaking countries ("DACH" area and more) and who are the people in that area, I realized that a German-language Mozilla planet would be a good idea and filed a bug for it.
• Various Discussions:
  My role in Mozilla, WebM and open video, Ratty's commit access (congrats!), "I believe" idea, buildbot 0.8, Mozilla network problems, commit access policy, tinderbox stats, missing MDC docs, etc.

The big things for me this week was surely the Data Manager, which I could bring up to show and manage almost all of the data I want to cover in its first incarnation. Version 0.3, which is up on AMO now, can display cookies, permissions, content prefs, passwords, and form data and also delete/manage selected entries of everything but the rarely used content prefs.
What's missing next to content pref deletion is sorting in some of the lists, some context menus on the lists, the "forget all data from this domain" function as well as dynamic updates of shown lists and tabs when the data changes from any indirect manipulation. Those are what I want to get done before calling this work a 1.0 version.
I'm also thinking about possibly replacing the tabs with sub-items in the domain tree - unless I find an even better idea. The tabs are good in principle but tabs within a tab feel somewhat strange. I'm open for good ideas there and in any other way on that Data Manager.
Still, I don't want to extend its functionality too much before going 1.0, as I'd like it to be integrated in SeaMonkey 2.1 and probably replace at least the Cookie, Image, Popup, and Password Manager windows there - consolidating all their functionality and making it easier to manage, I hope. After all, SeaMonkey should make it easy for the professional user to stay in control of his or her data.

Here's a summary of SeaMonkey/Mozilla-related work I've done in week 20/2010 (May 17 - 23, 2010):

• Releases:
  For SeaMonkey 2.1 Alpha 1, I did all I could to get what we could deliver in terms on release notes, wrote up an announcement, and finally got everything pushed live in the night from Tuesday to Wednesday.
  SeaMonkey 2.0.5 release notes should be finished, so everything's ready for a release in sync with the new Firefoxes on June 1.
• Build Infrastructure:
  As the large buildbot upgrade to 0.8 still needs to wait for a few issues to be fixed (Mozilla release engineering folks are testing it), I tried to improve the situation of master load problems by updating buildbot slaves to support log buffering, which should have helped somewhat at least.
• Build:
  My patch for startup crashes on updates has now also landed on branch, which should help esp. for downgrading actions in the future.
• Places:
  The places history speedup review moved forward and I attached new patches addressing comments. I hope this will be able to land soon. I also filed a bug and patch on fixing some of those issues on the Firefox side as well.
  I tried to find out when the places transaction changes on the Firefox side might land, as I'd like those and the history patch above to be done for my next places bookmarks patch set and try builds, but unfortunately, Mano isn't reachable right now.
• View Source:
  I've been bothering me for some time that SeaMonkey is using both toolkit's view source in some places, and it's own in others, when both are almost the same. When a blog post on its importance reminded me, I decided to look into using toolkit's view source consistently and overlaying our own features on it. I came up with a patch, though some toolkit fixes were needed to make this smoother. My patch is adding roughly 260 code lines but removing about 1650, so it's a clear net win in terms of code size, functionality is nearly unchanged, and all view source windows are consistent after the patch.
• Various Discussions:
  New add-ons manager, no Summit and my role in Mozilla, WebM and open video, OpenSearch, Ratty's commit access, facebook's plans to take over the Internet, gren/orange/tree-open stats, Sync brand confusion, Win64, etc.

As I found my role and the role of SeaMonkey questioned in the light of where the web and Mozilla are going, with large, often monolithic structures and companies casting their shadows and the suite, myself or possibly even the open web itself having a stain of being old-school, backwards, and conservative, I felt I needed to do something about it.
First, I decided to not keep so many things to myself but to blog more. And second, and even more, I once again became inspired by that agenda behind Mozilla, and decided to write up an open call to make a difference, pointing out why I think that Mozilla has unique potential and why we should beat the drum for our idea of the Internet.

I hope many people join and make a difference where they live, so in the end the open web and people's freedom win.

This Internet shall have a new birth of freedom - and that network of the people, by the people, for the people, shall not perish from the earth.

Here's a summary of SeaMonkey/Mozilla-related work I've done in week 19/2010 (May 10 - 16, 2010):

• Releases:
  SeaMonkey 2.0.5 moved forward, I readied just about everything except finalization of the release notes this week. Due to more delays with shipping Firefox 3.6.4, that has most of the same security fixes (in addition to the new feature that delays things) and us not wanting to ship either unexplained security fixes or explaining vulnerabilities the Firefox hasn't shipped an update to, we are now in a waiting position and will release as soon as they are, current plan seems to be June 1.
  For SeaMonkey 2.1 Alpha 1, we are not dependent on any other release as it's not a security update but a feature preview. Builds have been done and tested somewhat, no major issues have been reported, so we are on track for going public, which will probably happen tomorrow. Everything should be ready except release notes and the announcement, which both still need to be finished or written up.
• Build (and Win7 integration discussion):
  I did a few build system reviews, and followed the discussion in a bug about startup crashes on updates that more and more came to the conclusion that some component registry caching files we are dynamically creating should be removed on every update, which I finally made a patch for and could land already.
  Additionally, I filed a bug on a Windows installer build failure that happened after some mozilla-central changes to better support the Win7 taskbar, and jimm as the author of the platform and Firefox patches jumped in immediately and provided fixes as well as filed a tracking bug for Win7 taskbar integration on our side. He's been very helpful, and we appreciate that a lot. He even implemented platform support for per-window-type IDs, which we can use to optimize how mail vs. browser windows are treated by Win7. Help from you in the community to actually fill the gaps in our own code to make it all work would be appreciated, he surely would be willing to help there!
• Context Menu:
  You might remember that I had been doing work on a context menu test earlier. In that work, I realized we have bookmarking entries in there which need to be adapted for my places work, and that in turn made me realize it would be good to have a reference to the browser element the menu is being opened on. When I tried to implement that, I saw how alien indentation and code style in the file were, so the patch turned into a major cleanup of this JS file. I updated it this week for some review comments and could land it in the end, hopefully making this code better to work with. I also filed some followups to further improve this code.
• Places:
  I successfully ported the de-XBLification of bookmarks menu and toolbar to my local work on places bookmarks this week, but I still need to split it up and merge the work into the correct patch set parts. I might wait for review on places history speedup on our side or readiness/landing of places transaction changes on the Firefox side before I'll attach new patches and generate new try builds, though.
• Various Discussions:
  The name of MailNews, Account Manager, Summit, tree/network issues, new add-ons manager, release delays and OOPP blockers, Weave Sync L10n, Ratty's commit access, buildbot version upgrade, Thunderbird mailing lists without newsgroup mirrors, stage/FTP cleaning policy, etc.

Some time ago, the Czech Mozilla team asked me to answer a few questions about SeaMonkey - now this interview with me has been published (also available in Czech) - apparently they picked my statement how a marketing team could improve SeaMonkey usage as a headline.

Also, even while the first alpha is only about the be released, a few interesting things are in the pipe for or already landed in 2.1-targeted nightlies, like the new addons manager (see teaser screenshot (that still has a few things to be ironed out), our version of the findbar, customization of the mail search toolbar and more. And with your help, this already incomplete list can even grow more!

Here's a summary of SeaMonkey/Mozilla-related work I've done in week 18/2010 (May 3 - 9, 2010):

• Releases:
  I got 2.0.5 builds going (candidates are up for testing) and pushed them to the beta channel after some testing. Actual release might get delayed due to 3.6.4, with the same security fixes but even a large feature, is being delayed as well.
  For that release, I also went through L10n signoffs and after some review could add en-GB and zh-CN (Simplified Chinese) to our shipped locales - thanks to those L10n teams to provide this cool work!
  Also, I tried to get things lined up for cutting 2.1 Alpha 1, which took a while because of some blocker bugs needing to get out of the way (build breakage prohibiting Windows test runs, MDN breakage, Modern toolbar menubutton breakage). Meanwhile, into this week, it looks like we finally have what we need for this first Alpha.
  To support those releases and get them on track, I created preliminary release notes and filed a number of bugs tracking various stuff that needs to happen for the releases.
• Build:
  We had a very interesting build problem I looked into this week, which happened when the debug builds on Windows wanted to install tests for the content module. The error message of "nsinstall.exe: bad file number" was quite unhelpful, but when I investigated by taking one of our build slaves off the build bool and attacking it manually, I found out we actually seemed to be hitting a length limitation in the command line of the Windows build environment. We had been hitting this on those builds first (and not on other SeaMonkey or even on Firefox builds) as paths in that configuration are slightly longer than elsewhere. Installing 321 test files in one command line could make it grow even over 33,000 characters, and somewhere in that range we did hit a wall (smells so much like 32,767 but we seemed to get somewhat over that - maybe it's just argument string length or such).
  In the end, I made a patch for splitting that list of tests into two parts, the patch was accepted into mozilla-central and the problem is gone now.
• Places:
  The places bookmarks patch set has been on the backburner this week, but I followed as de-XBL of toolbars and menus landed for Firefox, will adopt this as soon as I come around to.
• Various Discussions:
  Account Manager, Summit, module owners, tree/network issues, etc.

I sent a lot of this week not only recovering from a cold but also at a local FOSS conference called "Linuxwochen" in Vienna. Lots of conversations there were pretty interesting, including those on getting an open geolocation database service going and thinking about organizing open source meetings together with the OpenOffic.org guys - oh, and probably even the one with the Microsoft guy who presented their "surface" desk and who didn't see why I disliked that it wasn't using an open software stack. The FSFE team congratulated me to that discussion the day after...
My talk on "Mozilla is more than Firefox" was quite well-received, people seemed to be particularly interested in mobile, Weave and Labs, and video, and someone also told me that "Mozilla should advertise the philosophy more". Also, the Mozilla swag I did get some time ago is now almost fully in the hands of the people.

Another fruitful conference experience with a lot of interesting chat with cool people - now let's put an interesting first 2.1 alpha into the hands of another set of cool people!

Here's a summary of SeaMonkey/Mozilla-related work I've done in week 17/2010 (April 26 - May 2, 2010):

• Context Menu:
  I did some more work on the context menu test and related accesskeys to fix Windows failures and cover some additional cases, and filed followups for selection tests and potentially hiding items.
  I had been stumbling over the "Bookmark This Page" and similar items of the context menu in earlier work, and now tried to fix that up to work correctly with places bookmarks, when I noticed I needed a reference to the browser element there. This led me to vastely re-styling nsContextMenu.js code to match current coding and whitespace styles.
• Bugzilla Cleanup:
  After a few weeks of discussion, I finally performed the proposed bug expiration - with the exception of help bugs - and switched almost exactly 2000 bugs in that move, leading to a massive decrease in bugs that need manual triage.
• Project Organization:
  Anticipating our Alpha 1 code freeze on May the 4th (today!), I poked responsible people on getting new-style flags for SeaMonkey 2.1 - thanks to Reed and Dave for their massive help there!
  Once we had the flags, I used them to triage blocking and wanted requests and ported all old 2.1a1 flag usages over to the new ones.
  In this last week before the mentioned freeze, we had a number of compile bustages from mozilla-central work, accidentally. I discussed those and tried to get someone to fix them, which succeeded in the end. Thanks to Bas, Standard8, and everyone else who helped there!
• Places:
  I worked on fixing a few bugs in the places bookmarks patch set, stumbling over the above-mentioned context menu thing there, and succeeding on getting rid of some, but I need to do more work there.
  Next to that, I closely followed some exciting work going on in places (de-XBL of toolbars and menus and modularizing transactions), which will need porting to our side once they land.
• German L10n:
  I did lots of updates for German L10n, bringing core and SeaMonkey trunk to green state for a short time.
• Various Discussions:
  Account Manager, Mnenhy, "site" vs. "web site" vs. "website", SeaMonkey 2.0 interview series, comm-1.9.2 branching, drumbeat catchphrases, hackers and posters, Summit, preparations for upcoming releases, new add-ons manager, etc.

I've been trying Account Manager this week, and interestingly it installed right away for SeaMonkey and even works somewhat! Of course, it's in a completely experimental state and doesn't do much yet as well as doesn't work with many sites right now, but that should change over time. Even when it still was "Weave Identity", I already found its ideas quite compelling, and I hope we'll be able to continue to have it working as well as possible with SeaMonkey even when it grows up as a Firefox project.

In other news related to Weave and SeaMonkey, I'm mentoring a Google Summer of Code project on Weave Sync engines for tabs and mailnews status for SeaMonkey this year, and had the first planning talks with Harini Sirisena on how we'll proceed with this. You can also follow Harini's blog to see what's going on in her project. Having your tabs from Firefox (desktop or mobile) or another SeaMonkey installations available on your SeaMonkey is surely nice (and almost there already anyhow), having the read status of feed entries and newsgroup messages as well as possibly message tags being synched up between installations would be really cool though and is the actual goal here.