The roads I take...

Here's a short summary of SeaMonkey/Mozilla-related work I've done in week 33/2011 (August 15 - 21, 2011):

• Mozilla work / crash-stats:
  Did the probably last per-build crash rate calculations for recent betas and compared them to the numbers that Socorro now generates directly.
  Worked with the Socorro team on some fixes to their recent release and filed some bugs and possible enhancements I came across.
  Blogged about new split crash-stats reports.
  As always, ran experimental reports on "explosive"/rising crashes and investigated interesting ones popping up in those.
• SeaMonkey Build/Admin:
  We ran into an interesting certificate problem with SeaMonkey updates - who would have thought that clamping the possibilities of issuers to one of the largest ones would cause a problem when that wouldn't issue any certificates any more? We'll need a very fast 2.3.1 release to get enough users of 2.1 and higher updated to a version that still will accept any updates in the future. Callek is working on that.
• Websites:
  I did an experimental implementation of a BrowserID login on my website system, still need to do more testing and some additional features like creating website accounts with it as well, but this looks promising.
  I also updated SeaMonkey Bug Radars for the rapid release cycle and they are mostly correct now, but apparently one more bug is left for the tracking ones.
  And on the SeaMonkey Development website as well, I added support for 2.6 and 2.7 version to ADU and downloads graphs in the metrics section.
• Themes:
  I updated EarlyBlue and LCARStrek for changes in the last two SeaMonkey and Firefox releases locally, but I still need to do some testing before I can release those.
• Various Discussions/Topics:
  Firefox version display and rapid release cycle, review lags, Web APIs, MPL2 RC, Upcoming phonebook, Mozilla website merge, more on MeeGo and Fennec, big mobile market news (Motorola, HP), etc.

I continue to be humbled by what awesome people work at Mozilla. Not just that we hired a lot of really great people (and we have a ton of additional career offerings but also about the quality of work that interns at Mozilla are doing. I always felt that it was great that we were having interns work on real-world projects that "normal" employees would work on as well, but I was completely blown away when I saw and heard some intern "show & tell" presentations about them implementing new language features for Rust, like object self-reference and inheritance. I mean, how cool is that for an internship work? They are creating stuff where I can hardly grasp their usage, let alone how an implementation could be done at all. I know, I have always seen people implementing programming languages as über-gurus, but people like that young man and young woman who presented doing that as a summer internship for Mozilla? I don't have a word for that - though, maybe, as it makes me actually stare in awe and wonder, there might be one matching the definition: purely "awesome".

Here's a short summary of SeaMonkey/Mozilla-related work I've done in week 32/2011 (August 8 - 14, 2011):

• Mozilla work / crash-stats:
  More per-build crash rate calculations for 6.0 betas.
  Collected more data for Flash long-time rate calculations, but probably need to do that report a bit differently.
  Looked into a sharp rise of Flash hangs, which turned out to be connected to some Zynga game, contacted their people on that.
  Had a ton of communication and discussion with the Socorro team to get their new, better beta/release reports and numbers just right, including me testing that work and giving feedback.
  Took part in a number of discussions on getting URL information for crashes.
  On trunk, I kept track of some fixes for recent parser crashes actually working.
  As always, looked into "explosive"/rising crashes with my experimental stats.
• Various Discussions/Topics:
  SeaMonkey 2.3 feedback/discussions, shipping "irreversible" changes in the rapid release model, review lags, Mozilla website merge, more on MeeGo and Fennec, etc.

This was a week where I concentrated forces together with the Socorro team to get the crash-stats server updated to support split reports for betas and releases in time before the next Firefox release hits, so we can avoid completely skewed statistics like those we had the last time around. We all knew it was a ton of work and we were cutting it close, but the team worked late hours and on the weekend, and on late Sunday they could make this release live, while the Firefox release is happening early Tuesday (tomorrow), which even gave us some buffer to make sure everything's alright - and it seems to have worked out! A big thanks and congrats to the Socorro team for getting it done!

I will blog tomorrow or so about more details on what's in there and how to deal with the new data that's coming in with this.

Here's a short summary of SeaMonkey/Mozilla-related work I've done in week 31/2011 (August 1 - 7, 2011):

• Mozilla work / crash-stats:
  Continued the per-build crash rate calculations for 6.0 betas.
  My reports on Flash hangs go on and have been augmented by more long-time rate calculations, but I need to do even more work there.
  I took more looks into confirming that our recent hang fix worked, which looks good, but also looked into more Flash-related issues worth a look, one of which an Adobe developer says they should fix for a future version.
  Again, I stayed in close communication with the Socorro team on their ongoing work for better beta/release reports and numbers.
  Blogged about crash rates after a recent change in Socorro went live.
  On trunk, I kept track of some fixes for recent parser crashes actually working.
  As always, looked into "explosive"/rising crashes with my experimental stats.
• Build System:
  Once again, I updated my patch for L10n-specific file removals on update, but I'm starting to give up hope that any of my mozilla-central patches will make it into the tree in foreseeable time - either my work is unwanted there or I'm to dumb to make it in a way that reviewers can swallow it. As I have more productive work to spend my time on as well, the motivation isn't really there to push on those much any more, in either case.
• German Planet:
  I pushed another design update done by Elchi3 live, thanks for the nice work!
  Also, I added Manuel Strehl to the aggregator, welcome to planet.mozilla.de!
• German L10n:
  I reviewed and pushed a number of small fixes to the German L10n overall, fixed some addressbook access keys (up to here affects all branches), updated DOM pieces of the L10n to current and renamed "Absturzmeldung" to "Absturzbericht" as that word matches better for crash reports that are not just notification-style but fully detailed reports.
  With that, quality of the German localization should be improved and all trees except Firefox and Fennec trunk (which don't lack much) are green for German now.
• Various Discussions/Topics:
  SeaMonkey 2.3 feedback/discussions, more on B2G and Web APIs, the way of features from prototypes to production, devtools, MIME library work, new UI concepts, community metrics, MeeGo and Fennec builds, etc.

The next round of releases is coming up and I think we're in pretty reasonable shape for them - the new release process is starting to sink in everywhere and that should make it a positive experience for everyone, as delivering more deep-going fixed as well as new features is easier for everyone, and the 12 weeks of stability testing without major changes that every release gets should vastly improve the chances of shipping a rock-solid product on the release channel from day one. While our team was not completely happy with the level of crashes in Firefox 4 and pushed for a 4.0.1 to fix them, we have better chances to spot such things and get them fixed in the beta phase already and that pays off.
I like what I'm seeing at Mozilla nowadays, but we still need to become better on communication and working together with the broader community - still, even that is being worked on.

Here's a short summary of SeaMonkey/Mozilla-related work I've done in week 30/2011 (July 25 - 31, 2011):

• Mozilla work / crash-stats:
  Did more per-build crash rate calculations for 6.0 betas.
  Also ran more reports on paired Flash hang details, hoping that Adobe folks might profit from those.
  Continued investigation of our top Flash hang on Beta/Aurora/Nightly and followed the regression fix that was found and deployed, tracking its success.
  Continued talks with the Socorro team on getting better beta/release reports and numbers, which we need for the next Firefox release.
  Also tracked their efforts on getting a release shipped that fixes "crashes per user" numbers to include plugin and content crashes.
  Investigated a number of new or rising crashes while some members of our team were out of the office, also noted two issues of concern in 6.0b3.
  As always, looked into "explosive"/rising crashes with my experimental stats.
• SeaMonkey Build/Releases:
  I closed up the first round of providing updates for linux64 builds by putting up the snippets for also offering the 2.0.14->2.2 "major update".
• (Tahoe) Data Manager:
  My patch for support of bare IPv6 addresses that also makes other cases more problem-proof got reviews and approvals and I landed it for all branches, including beta, so it's in SeaMonkey 2.3b2 already.
  I continued work on website storage support, but the test is giving me headaches, so I only attached a preliminary patch.
  Still, I did post a new version 1.4 on AMO that includes all this work and should be a significant jump forward for the add-on.
• SeaMonkey L10n:
  Reviewed and approved some sign-offs both for aurora and beta localizations.
• German L10n:
  Synched up German L10n of SeaMonkey with recent trunk changes.
• Various Discussions/Topics:
  More SeaMonkey 2.0.14->2.2 MU feedback, B2G, the way of features from prototypes to production, devtools, MIME library work, MeeGo and Fennec builds, etc.

The large topic this week was of course the B2G project which has been announced through newsgroups on Monday and I was called early on Tuesday to talk to a local media guy from "futurezone" about it - the resulting article (in German) contains comments from Mike Shaver, me, the announcements, and their opinions on it. In the end, I think this is a fantastic project for bringing web standards, including HTML, but also a number of other relatives, up to speed to gain the abilities to write really complete applications using them and to replace all apps usually needed on a tablet with them - from there, we can move to world domination for the web (or so).
Of course, a lot has to be done there, HTML needs to get good enough for real UI design, we need was to access more hardware functionality, but with proper security and privacy models, we need to improve offline support even more, make local and cloud services both equally usable for web apps, and probably more. Mozilla as the only non-profit player in the browser (or web application runtime) market is in the best position to work on those standards and implementations of them in a really open way that serves the user first, and that's why I think it's paramount that we play on this field.
The hardware and lower-level OS stack we start building prototypes with is only important right now in the sense of getting something to hack on and build this technology upon as fast as possible with as little work from our side as possible. Given that Mozilla has Tegras available in the office and a well-working Android browser, our team might as well kick off work based on those. Of course, I'd love us to use a more openness-friendly OS stack like MeeGo and a wider distribution of hardware in the future, but given the Mozilla codebase the whole work builds on, I'm sure that won't be a real problem, esp. if e.g. the MeeGo community helps out once code is there. Right now, the most important thing is to get code running as a prototype and the standards processes churning, everything else can be discussed later - we're far from even thinking about a mass-market or even small-market product in that effort anyhow. We can start on the the new web APIs right away, though, and that's one area Mozilla is already good at, so let's go for it and use that power to improve the web in an open way for everyone. That's what promoting openness, innovation and opportunity on the web is all about, right?