Hey folks! I'm in Ashland, Oregon today. :-) This coming week I'll be at OSCON.

I'll be giving two talks at FOSSCoach: Advanced Bugzilla, and Code Simplicity. They're both listed on the FOSSCoach Schedule.

I'll probably also be around the Mozilla booth when I'm not at FOSSCoach or at some OSCON session.

-Max

I'm planning a vacation in Washington State from August 14-24, back to Seattle and Vancouver. So if there are any Mozilla developers or fans in the Seattle or Portland, Oregon areas that want to meet up during that time, drop me a line in the comments section.

• August 14-17: Seattle
• August 17-20: Vancouver
• August 21-23: Seattle
• August 24: Back in San Jose

The first minor update to Mozilla Firefox 3 has been released. Firefox 3.0.1 fixes three critical security vulnerabilities, improves stability and resolves a handful of other small bugs.

The security fixes are detailed in the Firefox 3.0.1 section of the Security Advisories for Firefox 3.0 page. Two of the issues — one related to how Firefox handles command-line URLs to open multiple tabs and another allowing remote code execution by overflowing a CSS reference counter — were also present in Firefox 2 and fixed in Tuesday's Firefox 2.0.0.16 release. Security improvements in Firefox 3 mean that it's not vulnerable to some of the Firefox 2 variants of the command-line multiple tab exploit but it can still be compromised by combining the attack with a script injection flaw.

The final flaw only affects Mac OS X and allows an attacker to crash Firefox with a malformed GIF file, potentially gaining the ability to execute arbitrary code on the victim's computer. This vulnerability is not present in Firefox 2.

The non-security fixes include an issue where the phishing and malware database did not update on first launch and a problem that could cause Firefox to not save the security certificate exceptions list properly. A bug that could result in missing output when printing a selection from a page (bug 433373) was resolved and a Linux issue causing Firefox to always start in offline mode when using a PPP connection (bug 424626) was also fixed. The Public Suffix list has also been updated (bug 438585).

The Firefox 3.0.1 Release Notes have more details about the fixes in this minor upgrade. The new version can be downloaded from the Firefox product page or the Firefox 3.0.1 directory on releases.mozilla.org but most Firefox 3 users are expected to get it via the software update feature built in to the browser or their own operating system's update facility.

Talkback

Mozilla Firefox 2.0.0.16 was released this week. The stability and security update to Firefox 2 fixes two security bugs, which are detailed in the Firefox 2.0.0.16 section of the Security Advisories for Firefox 2.0 page. Both are rated Critical, the highest of the four ratings.

One flaw is related to how Firefox handles command-line URLs to open multiple tabs and allows an attacker to open potentially malicious URLs in Firefox from another application. One variant of this attack exploits the widely-reported Safari carpet-bombing vulnerability but others also exist. Somewhat ironically, the exploit relies on Firefox not being open at the time of the attack.

The other vulnerability allows an attacker to crash and run arbitrary code on a victim's computer by overflowing a CSS object reference counter. The detailed bug reports for both issues are currently access-restricted to avoid assisting attackers but will be fully opened after users have had some time to install Firefox 2.0.0.16.

Although Firefox 3 was released in June and all users are encouraged to upgrade, Firefox 2 will be maintained with security and stability upgrades until mid-December 2008, according to the Mozilla Developer News weblog, which reported on the release of Firefox 2.0.0.16 on Tuesday. Version 2.0.0.16 is the second Firefox 2 update to be released since the launch of Firefox 3 and follows on from Firefox 2.0.0.15, which fixed twelve security issues.

Existing Firefox 2 users will be offered 2.0.0.16 via the browser's built-in software update feature if enabled. It can also be downloaded from the older Firefox releases page or the Firefox 2.0.0.16 directory on releases.mozilla.org, where it is available in over forty localizations for Windows, Mac OS X and Linux. The Firefox 2.0.0.16 Release Notes contain more general information about the upgrade.

Talkback

The official Mozilla Blog has confirmation that Mozilla Firefox 3 now holds the official Guinness World Record for the largest number of software downloads in twenty-four hours. On Download Day, there were 8,002,530 downloads of Firefox 3 between 6:16pm UTC/GMT on Tuesday 17th June 2008 and 6:16pm UTC/GMT on Wednesday 17th June 2008.

Gareth Deaves, Records Manager for Internet and Technology at Guinness World Records, presented the official Guinness World Record certificate to Mozilla Europe President Tristan Nitot at a ceremony in London on Wednesday 9th July. While this official certificate is held by Mozilla, anyone who contributed to the record attempt can download their own personalized Firefox 3 Download Day certificate from Spread Firefox.

In a weblog post, Gervase Markham explains that the actual Download Day figure should be 8,002,529, as a Guinness World Records representative downloaded Firefox 3 during the twenty-four period, which cannot count as he is an official. Meanwhile, Mary Colvig has posted some details about the behind-the-scenes efforts that went into organizing Download Day and Asa Dotzler has published responses to some criticisms of the Guinness World Record attempt.

Talkback

Tomorrow, I’m comming to Moscow to meet with Mozillians
We’ll be in Rosie pub (google map) starting from 16:00 I think (at least me, because of my flight).

So, if you’re in Moscow these days, and have a free afternoon/evening tomorrow, and you think you might be a little bit thirsty - join us, I’d like to use this opportunity to meet as much of you as possible!!!

Then, on Monday, we’ll be going to Protva Open Source conference in Obninsk, where I’ll be presenting the vision Mozilla has on the Internet and the future of open web. It’ll be exciting for me to be there with all those people who work on the open source in the country that is currently doing a huge step toward open source schools, and if you combine with how big Russia is… amazing

So feel free to join us tomorrow or see you at the conference!

I didn’t realize it but both Ben and Taras will be at OSCON and will be talking about code analysis and code rewriting.

While that might sound boring, it’s not. In fact, if you’re interested at all in how to do analysis on a large C++ code base and see some of the fantastic tools that these guys have been building to help us manage the Mozilla code base you should come see this talk. Really. This stuff doesn’t really exist anywhere else in the industry and it probably has pretty wide appeal across a range of code bases - open source and proprietary alike.

June 27 (7.3; 195.1 total, 1978.9 to go; -7.7 from pace, -89.9 overall)

The next morning I wake up, finish off the second pound of strawberries and apple juice I’d purchased yesterday, add some photos to the first trail journal post. Incidentally, the laptop the hostel has for our use is running Firefox 2; I download and run the install program for Firefox 3 (released to the masses during my time hiking, and with a world record to its name, no less!) to find — gasp! — that the laptop is running that marvel of stability, Windows ME, while Firefox 3 requires Windows 2000/XP or newer. No upgrade for this machine! :-\ No Unicode love for the Firefox 2 users, either.

I then hit the trail with the Honeymooners at around 9:30 or so. I immediately confirm a fact I’d suspected yesterday: I bought way too much food when I did my shopping on an empty stomach. Best of all, we’re starting to hit the more mountainous part of Maine, and my pace becomes abysmal. Today I’d planned to hike to Spaulding Mountain Lean-to, a distance of 13.5 miles. Instead, I reach the Crocker Cirque (fun fact: the word cwm is defined by the official Scrabble dictionary as “cirque”; it is actually not a “Scrabble word” for me, because I’ve seen it labeling one of the features in the map in Jon Krakauer’s Into Thin Air) tent site, a grand total of 7.3 miles up the trail, of which the last mile or so is a rocky downward scramble — the absolute worst thing possible for my making any sort of speed.

The silver lining of this overfull pack, however, is that I eat very well today: two apples, three tortillas with peanut butter and honey, and a lot of gorp (with M&Ms added, naturally), and that’s just on the trail. (I’d planned to eat the bagels and apples anyway since they’re heavy, but I did eat them a bit more hurriedly than I’d expected.) Dinner is a Knorr chicken meal to which I add real chicken! Apparently, in addition to tuna in a pouch, there also exists chicken in a pouch! The miracles shall never cease. Mine is a lightly-garlic-flavored breast, although I hear they also sell unflavored breasts (not in Stratton, it seems). It goes pretty well, all told, and I sleep sated, again on a tent platform with a rigged-up staking system for my not-freestanding tent.

June 28 (6.2; 201.3 total, 1972.7 to go; -8.8 from pace, -98.7 overall)

The next day goes, if anything, worse than the previous day. I start the day pretty late, eating the remaining apples I’d started with and a few of the bagels I’d brought (which actually contain a pretty hefty number of calories, a fact I’d not noticed before in normal life). I’m still carrying far too much.

The trail is relatively uneventful. I pass South Branch Carrabassett River, a “ford” but in reality a bit of rock-hopping followed by walking an uncertainly-anchored plank across a ten-foot stretch of
water, around noon, finishing off the bag of dried fruit as trail mix that I’d started the previous day. (I picked it up from the hiker box at the hostel; it was amazing stuff but pretty heavy even after drying.)

The day ends just short of six in the evening at my target for yesterday: Spaulding Mountain Lean-to. I note in the trail register that the Honeymooners did stay last night here, and I make a note in my entry warning hikers not to buy groceries on an empty stomach.

The lean-to is a full house tonight: a couple on a three-day trip, it being the weekend and all, and a bevy of northbounders (five, the most I’ve seen yet in the same place). The northbounders are Crazy Legs, Water Bear, Marathon, Strides, and one last person whose name I’ve sadly forgotten. Dinner tonight is another modern miracle: beef Knorr with dried beef slices (again from a pouch). The slices are, of course, heavily salted; the instructions on the label suggest running the slices under warm water for fifteen seconds to wash off excess sodium, which I don’t do, much to my chagrin: this stuff’s wicked salty, nearly to the point of not being readily edible. The northbounders remark on the quantity of food I have (with good reason!) and say the bear canister I’ve been using on the trail is superfluous, with bear lines or boxes at every shelter (and perhaps tent site as well? I don’t remember the exact wording) where you’d actually need them. Given my pack weight now, I’m inclined to agree with just about any suggestion that reduces its weight.

Some fun facts about some of the northbounders from late-night flab-chewing:

One of them is from the upper peninsula of Michigan near Marquette (I believe the exact town mentioned was Big Bay, but I’m not 100% sure). He worked for the National Park Service on Isle Royale several years ago doing wildlife surveys, among other things, and once while hiking with a partner saw three gray wolves approaching him. He called to his partner hiking fifty or so feet ahead of him, and when the partner turned they’d all scattered. Since this was a bit unnerving, they quickly continued on for another mile or so before stopping to regroup. Then, on a whim, they checked whether any of the wolves had been radio-tagged: turns out one of them had been, and furthermore, it was only thirty or so yards away and had followed them the entire time. According to him, this was at the time the closest incident to a wolf attack in NPS history (!).

Another of the northbounders (perhaps the same one, perhaps a different one, perhaps the one who would then quite aptly have been named Marathon) has so far logged three forty-mile days on the trail during his thru-hike. I doubt I’ve broken fifteen more than a handful of times, ever, so this is an almost-incomprehensible feat. We shall see what flatter trails do to what I consider achievable.

June 29 (10.4; 211.7 total, 1962.3 to go; -4.6 from pace, -103.3 overall)

I get a nice quick start this morning in 75 minutes from in sleeping bag to hiking down the trail; I’m pretty sure it’s my fastest start yet out of a shelter (out of a tent is naturally slower), but I haven’t been paying much attention so far.

The first few miles are across reasonably flat ground, and I make nice time — for a moment I become optimistic that I can do well for the day. The first real stop is Poplar Ridge Lean-to at eight miles out, occupied by a couple guys drying out tents and other gear. I talk for a second to them (one has the same Eureka Solitaire tent I have), and they mention an under-construction tent site (not listed in books yet, of course) a little further down the trail between mountains — they inadvertently stayed there the previous night to be awoken by a trail crew coming in to work on the second of three days installing a privy. I then continue down the trail a touch and stop to use the shelter privy before heading along further; it’s too early and too short to stop here for the night. I pass over Saddleback Junior and stop at what I think is the nascent tent site, although it’s not marked by the orange tape that had been mentioned as denoting it. There’s a large number of boot prints from the south which stop right at the place where I turn off, so I think I’m there, even if I can’t find the aforementioned half-built privy.

The site itself is pretty nice; it’s not too buggy, although I pull out my rain coat to reduce the need to ward them off, set up my tent in a nice cushioned spot, and start making dinner. While making dinner, I hear some loud grunting noises from the east by the trail — turns out it’s a porcupine about ten or fifteen yards away! I don’t know for sure it was one since it was dusky, but it looked a lot like one; since I’m in the middle of dinner preparation the last thing on my mind is taking a picture — I’d rather he leave me (and my smellable goodies) alone now.

June 30 (10.4; 222.1 total, 1951.9 to go; -4.6 from pace, -103.3 overall)

Often trail magic, as it’s called, comes in edible form, a fact which should not be surprising given just how much energy hikers require to move from place to place (particularly through places like Maine which are particularly hilly). Today, trail magic was not edible but rather white and stored on a roll. More on that in a bit…

It turns out I’d stopped short of the nascent campsite, which I pass a bit down the trail this morning. Nobody’s there, and the privy looks complete, so sadly I can’t celebrate its existence; it looks pretty awesome. The morning continues with a nice hike up The Horn, yet another mountain in Maine. Saddleback Mountain itself follows next, and it has good views all around.

The most interesting part of the early day, however, isn’t what’s listed in the Companion. Strangely, it’s also not on any of the Maine maps I’ve seen. It’s definitely mentioned in a lot of books, but my available guide resources have both inexplicably missed it. It is a feature, located between The Horns and Saddleback Mountain, known as Saddle Point:

I fully expect the ALDHA and the Maine ATC to rectify this serious omission from their literature at the earliest time possible.

I continue down the trail toward the next lean-to, Piazza Rock Lean-to. Along the way I pass a guy wearing bug netting trimming along the trail; turns out he’s the lean-to’s caretaker out for a bit of maintenance. Since I’m hoping to make it six miles further that day to Little Swift River Pond Campsite, I continue on.

When I reach the lean-to, I decide to stop in to check when the Honeymooners passed through. Before doing so, I stop to use the privy. This privy has all the amenities:

Incredibly, while using the privy, I run out of toilet paper. I’d resupplied with a small partial roll in Stratton, but somehow, due to a ridiculously (but not perilously, as I note in the shelter register) active bladder, my supply (intended to reach Gorham) is already depleted! The options aren’t good. I’m right near a road into Rangeley, so I could hitch in in the morning just to get toilet paper, which would eat up a good hour, and that’s if I get lucky. I could experiment and see just how well leaves work, but I’d rather not find that out until I’m forced to do so. I also have one last option: see if the caretaker has a phone I could borrow to call for some to be delivered by friendly people in the area (there’s a nearby place with five people willing to put up and party with hikers that would probably work). I head back to the trail to think about how to find the caretaker without walking back too far, but amazingly he comes around the corner at exactly that moment! I make my request, and he goes one further and outright gives me a roll! It’s an absolutely monster roll that’s at least three inches in radius before flattening, but I don’t care: it’s saved me a lot of trouble.

Reinvigorated from this unexpected bit of magic, I complete the remaining 1.8 miles to the road in 48 minutes, at the strong pace of 2.25 miles an hour. If I keep that up for the next 4.8 miles I can make my intended destination with a little time to spare. Naturally, the uphills start again and I fail to do so, but I do make it two miles further down and set up camp just off the trail and next to South Pond, so water’s readily available. Bugs are out but not too bad as I make dinner and go to sleep.

July 1 (11.8; 233.9 total, 1940.1 to go; -3.2 from pace, -106.5 overall)

I’d planned to wake up around 6:30, thinking to pack up and be out before anyone else can pass by. Since I’m two miles from a road (and shuttles from town wouldn’t arrive until far later) and four miles from the nearest shelter, I don’t really expect anyone until at least eight. It’s really strange, then, when someone walks by at 6:35 as I’m slowly thinking about getting up and says hi. You meet all kinds on the trail, I guess.

After breakfast and packing up I head further south. I don’t go far before I pass a treasure: a Maine map for the section of trail I’m hiking! The Maine topos are pretty awesome: profiles, notes on trail features passed along the way, and so on. I haven’t bought them because they’re expensive, and many thru-hikers have done fine with trail signs combined with the white blazes on trees and rocks along the way. This map is only good another few miles, but I’ll take what I can get. It’s probably the early hiker’s map, but if it isn’t, I’ve got a nice shiny map (and if it is, he’ll be happy to get it).

A few hundred yards later I stumble on an even better treasure: the map for the next section of trail. Again, it’s probably the early hiker’s map, but whatever: I have far-seeing guidance for a little while.

Later down the trail I meet another hiker heading north named Saxifrage (a thoroughly awesome name). After inquiring where he’s heading to make sure he’s not doubling back for maps (he’s not), he mentions that a guy up the trail, a Seth of the American Hiking Society, had asked him to look for the maps. I tell him I picked up the maps and will pass them along when I see Seth.

I eventually catch up to Seth at Sabbath Day Pond Lean-to, where I eat lunch and pass on the maps. He’s delighted and passes along contact info in case I need any trail magic in the D.C. area; we’ll see how I feel when I get there.

I’m hoping to head on to the next lean-to, Bemis Mountain Lean-to, so I hurry on, past an awfully large number of hikers — I think I pass around twenty total today (twelve or so from a camp group). Before getting to the lean-to I have a road crossing and a ford, not to mention the mountain itself. The road crossing is a definite candidate for best road-crossing view; I’ll post the picture when I have more time later.

The descent from crossing to ford is steep and takes a lot of time, so I decide to stop just after the ford and pitch the tent. This isn’t actually a good idea: the bugs are the worst they’ve been since the one site in the 100-Mile Wilderness, and I pull out the raincoat and forego a cooked dinner in favor of energy bars and call it a night.

July 2 (12.5; 246.4 total, 1927.6 to go; -2.5 from pace, -109.0 overall)

I get a late start in the morning; bugs are out again for breakfast, so I pack up quickly and don’t heat water for oatmeal. First stop is the lean-to I’d intended to hit the previous day, after a long hike over Bemis Mountain. I see I’m still behind the Honeymooners, and I discover that the two hikers with whom I’d hiked Katahdin oh-so-long-ago are now known as Asgask and Slowpoke. I sign the shelter register and include the birthday song in my entry, because it’s my birthday today! Happy birthday to me! (22, a pretty unremarkable number, but still a new age.)

Travel is fine until just before Old Blue Mountain at about ten miles in. I hope to go seven miles further to a shelter, but just then it starts to rain pretty hard, and even with a rain coat I’m completely soaked when I reach the top of the mountain. Since I’m just short of a road to Andover, I decide to stop there and hitch in to a hostel for the night — a birthday present, or something. The only problem is that the Companion says of the two roads into Andover that, “Neither road has much traffic.” I hurry as fast as I can and hope for the best.

I reach the road at 20:15, at which point it’s not yet dark but is starting to get there. One vehicle passes heading in the opposite direction as I descend the last few feet to the road; I then wait fifteen more minutes, during which time exactly two vehicles pass heading in the direction I want to go. Miraculously (and I mean that in the literal sense) the second vehicle offers me a ride, next to their two dogs in the back seat of their pickup truck. They take me nine miles into town, and I arrive at the Andover Guest House at 20:52. I say hi to the hostel owners, who tell me I probably want to head over to the store across the street to get something to eat — just before it closes in eight minutes. (See what I meant about miraculous? You can’t cut it any closer than that!) I get food and eat it while talking to family back home on the phone; when the conversation ends I check email and deal with a post here while laundry cycles, then I head to sleep.

The main theme of southern Maine is that it’s a lot of up and down. Northbounders are blowing past us, fresh from 2000 miles of hiking and the White Mountains as conditioning for the ups and downs (south of the Whites is actually considered pretty easy — the “hiker superhighway”, actually, where twenty-mile days are neither uncommon nor generally difficult if you get up early and keep a nice, long stride going), so their pace can be a bit depressing if you don’t take it in context. Right now I’m just anticipating Glencliff in New Hampshire, where, as I’ve been told by the two hikers at the lean-to just before Saddleback, the mountain climbing ends and the hiking starts.

I just pushed version 0.5.0 of the SHOUTcast Directory add-on for Songbird which adds a preference feature for allowing users to type in a comma-delimited list of genres. Interestingly, I knew this would be a requested feature when we pushed 0.4.8 in conjunction with Songbird 0.6, but I didn’t know how much. Turns out a lot of international folks wanted to be able to list different language genres, and that this missing feature was a lot more popular than I expected.

Anyway, version 0.5.0 supports this now. Time to get back to my OSCON prep…

Thanks to the heroic efforts of our GStreamer developer, we’re bringing partial GStreamer support into Songbird far ahead of schedule.  Those of you on Windows and Mac (i686 only) will now be using GStreamer to play back your FLAC files.  This change is purely under the hood, so we don’t expect you to notice anything more than a smoother FLAC playback experience.

We also have a little secret about the GStreamer integration in our next release — it’s fully integrated for all supported file types.  We’ve turned it off by default to give our community a chance to tinker with it for a little while before pushing it out fully (it will take one more release before we’re ready to fully switch over to GStreamer as our media core).  If you want to be one of our bleeding-edge, 100% GStreamer mavericks, you’ll need to set one environment variable.  There’s more info on that here.

Grab your GStreamer-enabled nightly build here (all standard caveats about nightly builds still apply).  As always, we love to hear your feedback.  Please let us know how GStreamer works out for you!

I’ve been working on something fun for the last couple of days (in between patch-revs on other bugs I can’t talk about yet), and I thought I’d blog about it. It’s cool (at least to me) and useful:

John Resig and others involved in the standardization process for the new ECMAScript language proposals have been going back and forth lately over the design of a new language library feature, to facilitate copying properties of one object to another. Object.extend() — essentially you pass an object (the one you mean to extend), and a series of objects (from which you want properties copied), and you get a new object with all the goodness of the others:

From the spidermonkey JS shell:
js> var a = { get x () { print("foo"); return 3 } }
js> var b = [ 1, 2, 3 ]
js> var c = { foo: “bar” }
js> var result = Object.extend({}, a, b, c)
js> result.toSource()
({get x () {print(”foo”);return 3;}, 0:1, 1:2, 2:3, foo:”bar”})
Notice the getter itself is copied, not the result of its evaluation.

I hope this feature makes it into the standard(s), I think it’s a very powerful, useful, and flexible one. So much so that many of the existing AJAX libraries already have a 5-or-10 line version of it included, under any of a number of names, and with a few variations in semantics. In my opinion, a number of the other features being discussed in lieu of it (like Object.clone) can be implemented trivially in terms of it). A good, consistent semantic for a feature this convenient seems like a great thing for the “library” side of JS, to me. I’d love to land this for 1.9.1, it seems like it would, at least, be handy for extension authors. I hope it survives the committees!

Also, thanks to John for writing a nice test-suite for the feature.

I’m going to be at OSCON next week. Taras and I will be hosting a BoF session on using automatic analysis and rewriting tools for open-source projects on Wednesday evening. You’re welcome to come and learn about our tools, watch demos, or just just heckle and meet!

I’ve been to OSCON once before. I got a chance to meet people in person I only knew via email, and also meet some new people who had read my blog or otherwise knew of me. It was a blast. I hope to meet even more people this time around. Many of the official sessions don’t look that exciting to me, so I might spend a decent amount of time at OSCamp or just talking to interesting people. If you’d like to meet me and can’t make the BoF session, you can probably catch me at the Mozilla booth.

I’ll be staying the following weekend in Portland, so if you have recommendations of restaurants or sights that I shouldn’t miss while I’m there, comments welcome.

Platform/2008-07-16 From MozillaWiki

« previous week | index | next week »

Notices

• 1.9.1 Alpha 1 Freeze on Tuesday, July 21st @11:59PM PDT.

GFX 1.9.1 Update

• GFX blocking 1.9.1+

• GFX wanted 1.9.1+
• borders
  • optimizations landing this week

  • border-image needs some fixes, landing soon
• SVG fonts — no news
• Do we have an update on bz’s work to make <img src=”foo.svg> possible?
  • will be done via bling-branch
• downloadable fonts (jdaggett/zack)
  • Waiting on reviews, might happen by alpha, might not (john’s out this week)
• pixman/cairo perf:
  • new cairo drop this week

  • jeff finding big perf wins in pixman C paths
• CMS perf work:
  • down to 10% perf hit from 20%, bobby has plans for the remaining 10%
• UI improvements:
  • Glass blocked on widget bug (robarnold)
• imglib:
  • got decent results with the patch for bug 430061 - tryserver builds available

Layout 1.9.1 Update

• 1.9.1 Layout Bugs

• Video (roc/cdouble)
• Acid3: Roc or dbaron, status?
• SMIL (dholbert):
  • Implemented getOverrideStyle

  • Working on building SMIL-for-CSS-properties on top of that
• SVG fonts:
• SVG CSS, roc:
• border-image (robarnold, dbaron):
• media queries (dbaron):
• CSS transforms - keith:
  • Click detection and event coordinate space translation working.

  • Working on a few stray bugs.

Content 1.9.1 Update

• Content 1.9.1 Bugs
  • Worker Threads

  • Cross-site XHR:
  • Java/OJI bug:
  • Selectors API (Patch ready, just needs review)

JS 1.9.1

• JS 1.9.1 Bugs

• JS Tracing:
  • Last week’s status:
    • more nanojit patches

    • added tracing for many opcodes, close to running some benchmarks unmodified
    • perf looking good
  • This Week’s Status:
    • Some unmodified benchmarks

    • many more opcodes traced
    • perf still looks good

Some new wanted+ bugs. In particular, we have an ES3.1 Decimal patch + tests from Sam Ruby.

Adding profiling hooks for other platforms in the spirit of our Shark JS hooks. Graydon Hoare has landed support for Callgrind, and a VTune patch awaits review.

Blocking
417131 igor@mir2.org review_brendan? JS Enumeration Allocation Consternation
430133 mrbkap@gmail.com review_brendan? Object.defineProperty
421864 igor@mir2.org wip-patch Interpreter creates too many doubles
433337 igor@mir2.org checkin_needed Reunify jsinterp.c on Windows
Wanted
229756 brendan@mozilla.org Make SpiderMonkey’s const extension JS2/ES4 compatible
260106 brendan@mozilla.org wip-patch elisions in array literals should not create properties (js1_5/Array/11.1.4.js)
312354 brendan@mozilla.org FIXED Assignment expressions have wrong type (ecma_3/Operators/11.13.1-002.js)
363534 brendan@mozilla.org FIXED Combine JSOP_LT and JSOP_IFEQ, etc., pairs
384244 crowder@fiverocks.com review_igor? update jsdtoa with interesting pieces of more-recent dtoa
433351 crowder@fiverocks.com Implement Object.extend
442379 dmandelin@mozilla.com wip-patch try inline-threading, at least with GCC if not MSVC
305064 general@js.bugs Add trim, ltrim, and rtrim features for javascript strings
352437 general@js.bugs string.link does not escape url
429507 general@js.bugs Function.prototype.bind
411575 igor@mir2.org FIXED js_PutCallObject() is slow.
432881 igor@mir2.org FIXED SM: JSVAL_VOID as a pseudo-boolean
443746 igor@mir2.org FIXED Optimizing the enumeration state allocation
346749 mrbkap@gmail.com still_want? let declarations at top level are turned into var declarations
442358 mrbkap@gmail.com FIXED jsinterp control flow doesn’t flow as specified
312116 nobody@mozilla.org should support catchall getters/setters
445178 rubys@intertwingly.net Decimal Support
419225 sayrer@gmail.com wip-patch refactor ExecuteREBytecode and SimpleMatch
430930 sayrer@gmail.com Date.parse cannot even parse “2008-04-26″ (should understand ISO 8601)
419743 shaver@mozilla.org wip-patch JSOP_CONCATN for improved chained-concat performance
433335 shaver@mozilla.org mark sharp object avoidance
433336 shaver@mozilla.org array iteration optimization

Mobile 1.9.1 Update

Security

• bug 331088basically untouched since March

• bug 371370last touched in January

• bug 413048last touched in April

• bug 422283last touched in April

General 1.9.1 Updates

• offline, dcamp:
  • Last week:

  • This week:

mozilla-central Retrieved from “http://wiki.mozilla.org/Platform/2008-07-16“

We are proud to present the winning screencast made by the one and only, Cameron Roy:

• Article: How to customize the toolbar
• Screencast: Click to view

His video won because it best satisfied the four judging criteria.  For one, Cameron showed how to customize the toolbar, which is not the easiest concept to communicate or digest.  For many users, it’s a new concept to change around the position of the buttons that sit in the browser.  With his great video, Cameron really demystified how challenging this could have been for an end user.  The production quality met our standards with great timing and flow (not too fast, not too slow) and everything fitting well in the screencast.  Of course, we would not have chosen him if his video had not covered the concepts completely.  Finally, his screencast really matched the information in the article.  Users who read this article and view his screencast will have a very detailed explanation with an illustration of just how to rearrange the buttons on the toolbar.

This is exactly the type of screencast we were looking for - a visual medium to help translate and communicate technical concepts to a wider audience. Overall, a terrific effort by Cameron.  Nice work!!

The other winning screencasts will be announced soon…

It’s time for another Monthly Meetup. This month’s meetup will be next Thursday, July 24th, 6pm at Mozilla’s office - 1981 Landings Drive, bldg K in Mountain View, California.

There will be informal lightning talks and progress updates on the various active Labs projects as well as plenty of opportunity for discussion and hacking.  We will be streaming the evening out to the Labs site.

If you are in the Bay Area we’d love to see you next week!  Please take a moment to RSVP by commenting on this blog. We don’t want to run out of pizza again!

Havoc mentioned that he had talked to me about the emerging GNOME duality after coming back from GUADEC. He also suggested that I post something about it. I’ll try to put my thoughts down here on paper.

First of all, I think that the GNOME project is in a really strange place right now. Others called it “decadence” but I’ve been thinking of it more as a state of transition. Things are changing in GNOME and I think that it’s largely just a reflection of time. GNOME has been around for more than a decade. Leadership is changing. GNOME is evolving into something else. I don’t think that I really understood this until after I was at the advisory board meeting at this year’s GUADEC.

Second, I’ll state what I said to Havoc. That I think that GNOME has evolved into two different projects, each struggling to share code and be successful. One is the “classic desktop” as we used to call it at Red Hat and the second project is built around servicing the mobile and highly-specialized desktop experience market. These are not the same thing, not by any stretch. Different audiences, different goals, different players and different revenue models. As a result you can feel some measure of friendly, but sometimes misunderstood confusion in the GNOME project because the underlying change isn’t fully understood.

A side effect of this change is also that there’s been a major change of the leadership in GNOME. It’s safe to say that in the past the North American Linux distributors played the largest part in determining the direction of GNOME. But those two companies haven’t experienced much growth over the last 10 years, either in terms of market or financial success and that’s been reflected in the relative investment that you see from the two companies.

Novell has lost most of the former Ximian people to attrition after the Novell merger and Red Hat hasn’t invested in a huge number of people relative to the growth of the rest of the company. And desktop opportunities haven’t really grown so things are at best static. Novell is still invested in apps and Red Hat is still invested in underlying infrastructure. Not much has changed on that front in a long time. It’s still all slow-growth high-touch enterprise work. Not the kind of stuff that shapes an industry, to be sure. (Note that I don’t mention Canonical who intentionally does not invest in upstream GNOME and aside from a somewhat large distribution channel, does not lead inside of GNOME.)

This is in direct contrast to the various consulting companies doing mobile work. There are at least 14 20+ people at OH, 20+ people at Collabora, 11 people at Imendio, people at Fluendo, people at Nokia, Litl, a couple of other random consulting shops in places like Spain, all leading and all contributing. These little shops are alive with energy and impact, they are writing new and interesting things and they are really the heart of where GNOME is heading. This is where the new leadership lies. And they are managing to lead as a group and share in the wealth found in GNOME work.

As a side note this signals a change in the geography of leadership of GNOME. It’s moved from the US to Europe in a very short time frame. If you want to be part of the core of GNOME you have to be in Europe. It used to be Boston in the US. That is no longer the case.

But this also signals a different kind of change. An even more important one. In the past the core of GNOME has been creating great experiences for users. GNOME love. Fixing hard technical problems up and down the stack that really solve user problems. D-Bus, HAL, PackageKit, NetworkManager - all of these low level pieces started with creating great user experiences. This is what has differentiated GNOME from KDE - creating simple and wonderful experiences for users. It was never technology focused.

But in the mobile world, and certainly one that’s driven by the financials of consulting organizations, platform is key. (Actually, this is a lie, but bear with me for a second.) The new leadership is more interested in delivering a working platform than a great experience. The experience in the mobile market is more often determined by the customer. In this case it would be the manufacturer or the mobile operator. And that’s something that GNOME has never had to deal with before, at least not in any real sense.

I will make a statement here that will probably be controversial but I believe to be true. Based on experiences with developers, GNOME’s stack is not as good as others that are on the market when it comes to platform-based decisions and platform experience. In particular, Qt and Apple’s platform are both very developer friendly and make it very easy for developers to create good-looking, modern looking apps. And they have more time and experience in embedded environments. With GTK+ it’s possible to make things look good and work fast, but it’s very difficult. If in a platform-driven market and a platform-driven world you’re not the #1 or #2 player it’s going to be very difficult to make a dent in the market. (This is especially true if Nokia decides to fix the Qt licensing.)

Basically what I’m saying is that if mobile is one of the main roads for GNOME (and it clearly is) then platform and embedding needs to be a core competency of the entire project. It can’t be just about experience. And I don’t think that we’re there today. That’s one of the big risks that we face.

I made a statement earlier, one that’s actually a lie. That platform was the key component in the mobile space. But I would actually argue that it’s not. End user experience, which can be driven by the platform in a bottom up fashion (as opposed to top down) is extremely important. Apple showed us that. The ability to make decisions about user experience in mobile is something that hasn’t historically been up to platform people. But I believe that in order to really be successful that GNOME needs to find a way to do this.

I believe that this is also one of the things that is a big downside of having a lot of small consulting shops all working on the same project. While it’s nice to have that diversity, it’s also very difficult to come to decisions about experience. And no one right now is focused on a complete product and experience. Everyone is building tools to let someone else make a decision.

Products almost always win. They lower costs for buyers and producers alike and increase leverage for the people who build the products. They offer the producer the ability to make decisions. And that’s the most important factor in building great experiences for people. Weak products are the result of poor decision making and a lack of design direction. And I worry that’s where GNOME is going. To a place where we’re not at our core competency and we aren’t building things that actual people will care about. That’s not historically what GNOME has been about, and I worry that’s where it’s going.

That’s also why I’m calling this a duality. Two projects existing as one. One without a great deal of success, but an end-user identity and brand and goals, and another with a chance to succeed but without much identity or end-user goals. It’s going to be a rough ride and I think that in order to find success we have to find a way to merge the two into a single set of goals. The desktop isn’t going anywhere, but the mobile project isn’t going to produce GNOME-visible results either. GNOME will be well-hidden behind someone else’s branding and experience. And maybe that’s fine, but it’s not the way to lead and win in the end.

I think the GTK+ 3.0 discussions are a symptom of this. They are sold as a technology change, a chance to clean up, but without a specific user-driven direction as to why someone would want to make that change. (Or at least that’s how it was presented at the AB meeting.) If someone could describe why that change should be made, that would be great. How will it help users? How will it help us achieve our goals? (Holy crap, what are our goals anyway?) How does it help bridge the gap between desktop and mobile? These are important questions, and I think a lot more discussion needs to take place.

So those are my thoughts on the GNOME project today. Hope it helps bring some understanding to how GNOME can more effectively manage itself.

After a couple of late nights reloading repository dumps and learning new and wonderful unexpected “features” of mixing svn:externals and svnsync, I’m happy to announce that publicsvn.songbirdnest.com is finally available again.

Timelines are back up as well.

All external developers will need to repull their trees; these repositories—multiple—are entirely new SVN repos, with different UUIDs. Some Subversion clients will ignore this (minor?) detail, and you can still update repos, but it’s likely to cause grief later.

Also, the repository paths have changed; please see our developer documentation on Checking out the Code for details on the changes and to figure out what you will want to repull.

We apologize for the inconvenience caused by having to repull trees, but we were long due for a repository reorganization.

An overview of some of the changes:

• Different kinds of code are split out into different repositories
• Branching and tagging will now follow standard Subversion practice, i.e. no more $tagName/trunk directories.
• Brand new, faster publicsvn server (thanks Tyler!)
• https is no longer supported on publicsvn, since it’s not necessary. This reduces load on the server, meaning faster pulls for you!
• Backups are more manageable; backups of our main repository now take five minutes instead of three hours; restores take twenty minutes instead of twelve hours); this means other processes, like mirroring changesets to publicsvn, is faster!
• We’ve pruned code that wasn’t necessary for a standard Songbird build, meaning faster pull and update times!

Thanks to everyone for being patient while we worked all the details out, and thank you to Mitch on IRC, whose pestering kept me honest while we did this; I needed it.

In my original post on whoisi I made a remark about wanting to keep track of events:

You also might notice that I have an entry above that just says “@fisl2008″. This is me just playing with events. One thing I’ve always wanted is the ability to say “I’m going to be at this event and I would love to see others who are doing the same.” In this sense it’s like saying “I was @ FISL 2008.”

Over the last couple of evenings I put together some code that made that possible and decided to make a stream for OSCON 2008 since I will be there:

I go to a few conferences a year and I’ve always wanted to have access to information about who was there and what they were doing during the conference. So I’ve tried to create that experience with this. Want to keep up with who is going to dinner where? Or who found something really interesting? Following this stream should help.

If you look at the events page there’s a quick tutorial on how to add yourself or someone else to the event. It’s easy. And thanks to a suggestion from Jeff the aliases in profiles now link to the respective event and/or group. That should make it easier to discover events from people’s pages.

There aren’t that many people listed right now. About 34 last time I counted. That’s about the number of people who responded to my request the other day. But if you’re interested in the feed or you’ll be there feel free to add yourself.

Have fun!

Just a reminder…. we’re going to be at OSCON next week! If you’re going to be in Portland, make sure you come to Beerforge, Thursday night (the party we’re co-hosting with a bunch of other rad companies).

Come stop by our booth (we’re hanging out with our favourite peeps from the Mozilla Galaxy) and buy our super-exclusive OSCON-only “Open Your Mind” t-shirt. We’ll have a limited supply of our other shirts too if you want to get your bird on. All proceeds are being donated to the Mozilla Foundation so you can feel good while you look good. We’ll have a bunch of stickers, pins, posters as well so you can sticker up your laptop… or your forehead, whichever you prefer.

And last but not least, stop by my talk Thursday afternoon @ 14:35 in E146 and say hi!

Okay, most programmers know the story—way back when, somebody found an actual insect inside a computer that was causing a problem. (Actually, apparently engineers have been calling problems “bugs” since earlier than that, but that story is fun.)

But really, when we say “bug” what exactly do we mean?

Here’s the precise definition of what constitutes a bug. Either:

1. The program did not behave according to the programmer’s intentions.
   or
   
2. The programmer’s intentions did not fulfill common and reasonable user expectations.

So usually, as long as the program is doing what the programmer intended it to do, it’s working correctly. Sometimes what the programmer intended it to do is totally surprising to a user and causes him some problem, so that’s a bug.

Anything else is a new feature. That is, if the program does exactly what was intended in exactly the expected fashion, but it doesn’t do enough, that means it needs a new “feature.” That’s the difference between the definition of “feature” and “bug.”

Note that hardware can have bugs too. The programmer’s intention is rarely “the computer now explodes.” So if the programmer writes a program and the computer explodes, that’s probably a bug in the hardware. There can be other, less dramatic bugs in the hardware, too.

Essentially, anything that causes the programmer’s intentions to not be fully carried out can be considered a bug, unless the programmer is trying to make the computer do something it wasn’t designed to do. For example, if the programmer tells the computer “take over the world” and it wasn’t designed to be able to take over the world, then the computer would need a new “take over the world” feature. That wouldn’t be a bug.

With hardware, you also have to think about the hardware designer’s intentions, and common and reasonable programmer expectations. At that level, software programmers are actually the main “users”, and hardware designers are the people whose intentions we care about. (Of course, we also care about the normal user’s expectations, especially for hardware that users interact with directly like printers, monitors, keyboards, etc.)

-Max

Comments: 0