No idea the cause. happens frequently using vmware fusion on a macbook pro running ubuntu.

Four years ago, i wrote a bit of code that did directional navigation in the browser.  Basically, if you press an arrow key, we would move in that direction.

I wrote this first implementation in C++ and totaled about 2300 lines of code (about 100 lines were license headers).   Although lines of source code isn’t a great indicator of complexity, I hope you can trust me that the code is, hmm., not very easy to follow at times.

We want to have similar functionally in Fennec.  So, today I rewrote the code in Javascript.  It is only 360 lines — 15% of the size! (about 30 lines are license cruft.)  It isn’t a complete implementation, but does handle most of the cases.  Even if the Javascript code doubles, it will still be easier to understand.

Wait, there’s more –  it performance seems as good as the native implementation on most pages.  Four years ago, writing anything like this in Javascript would have not been an option.  Today, the opposite is almost the case.  Simply amazing stuff.

Aza posted a mobile ui mock up via a screencast which you should check out.  This is similar to the other ui proposals, but it goes alot further.  Engadget picked up the blog post which has some nice comments.

Reflow is the process by which the geometry of the layout engine’s formatting objects are computed.  Last week, I was searching around and stumbled on these really cool visuals of what a reflow looks like.  If you’re a visual person, you will dig these:

Reflow of Mozilla.org:

Reflow of Google.co.jp:

Reflow of Wikipedia:

If you produced these, kudos!  If you know how to produce such things, please drop a note!

Update:

Satoshi Ueyama is the author of this cool demo.  He sent Gen a note on how he did it:

First, he builds Mozilla so that the layout portions (nsIFrame::SetRect, nsIFrame::SetSize, nsIFrame;;SetPosition, PresShell::DoReflow, etc) are set to output to the log file.

He then gives an example of the log of each of the frame movements from Reflow:

{"event": "MoveFrame", "frame": {"address": "0x04FE41A8", "type":
"TextFrame", "content": {"address": "0x05014150", "type": "other"},
"rect": {"x": 0, "y": 0, "w": 0, "h": 0}}},
{"event": "MoveFrame", "frame": {"address": "0x04FE41A8", "type":
"TextFrame", "content": {"address": "0x05014150", "type": "other"},
"rect": {"x": 0, "y": 0, "w": 0, "h": 0}}},

Here is an example of dump from the termination of the tree structure.

{"event": "FinishReflow", frame:
 {"address": "0x04FE3E28", "type": "ViewportFrame", "content": null,
"children":[
     {"address": "0x04FE4000", "type": "ScrollFrame", "content":
{"address": "0x04FD4858", "type": "element", "tagName": "HTML"},
"children":[
         {"address": "0x04FE43EC", "type": "ScrollbarFrame",
"content": {"address": "0x04FFEFF8", "type": "element", "tagName":
"scrollbar"}, "children":[
             {"address": "0x04FE4578", "type": "BoxFrame", "content":
{"address": "0x04FFE0F8", "type": "element", "tagName":
"xul:scrollbarbutton"} }
     ,
       {"address": "0x04FE47BC", "type": "sliderFrame", "content":
{"address": "0x04FFE278", "type": "element", "tagName": "xul:slider"},
"children":[
                 {"address": "0x05004E6C", "type": "BoxFrame",
"content": {"address": "0x050030E8", "type": "element", "tagName":
"xul:thumb"}, "children":[
                     {"address": "0x05004FDC", "type": "BoxFrame",
"content": {"address": "0x050013F0", "type": "element", "tagName":
"xul:gripper"} }
         ] }

He then makes the video animation (for Google Video he suggests Ruby+Ruby/SDL, for the video from Mozilla 24 he used C+++Direct3d.) As memory address is outputted to the log each frame can be distinct.

The behavior of each frame in chronological order is reproducible as the log has each frame’s memory. (For instance, in tracking the frame “address”:”0×04FE41A8″, you can reproduce that frame’s behavior chronologically.

I created a patch that allows people to change the default buffer sizes that our networking layer uses when loading content from the web.  The hope here is not so much to get this patch into the tree (it would be nice, but the patch isn’t ready for primetime), but rather to ask you to take a look at the build with this feature, tweak the values, and send me some feedback.  Be sure to take a look at the comments in the bug, specifically concerns about a balance between user perceived performance and clock time performance.  Simply taking the values that result in the fastest page loading (clock time) may not be the right values for human beings watching the content be loaded.

Here are the builds.  (these are experimental builds).

To change the values, edit all.js:

pref("network.default.buffer.size", 16384);
pref("network.default.buffer.count",32);

The defaults in Firefox are 4096 and 16 respectfully. You probably can change these values via about:config (and the patch doesn’t prevent you), but that may be bad.

I added a new attribute on the MouseEvent to expose the amount of pressure is being applied to a touch screen.

The video shows me pressing a pen then my finger on the touch screen.  The value on screen is a number between 0 and 1; zero indicating no pressure and one indicating alot of pressure.  Sorry about the quality.

You could imagine building a web painting application that varies the thickness of a line based on how hard you press, or the Fennec browser app could vary the speed of the scroll based on how hard the user presses on the scroll area.

Before this makes it into the project, there are a bunch of unanswered questions in the bug.  Mostly questions relate to how we expose this property in the DOM, and how to gather this pressure information from the platforms.

Chris Blizzard has a great summary of our last embedding meetup.  It details the direction we are taking with “how to embed mozilla in your application”.

We are having our next meet up on May 8 and 9. If you are interested in participating, please rsvp with blizzard.

Mitchell Baker spoke at Web2.0 yesterday.

Some of the highlights that I wrote down when I watched it (paraphrased):

“For most people the mobile web should be irrelevant…. there is only one web.”

Exactly. Wap is long dead, long live the web. (if you are still shipping wap content, the world is no longer flat, rejoice!) The point is that I as a user should be able to go to “wellsfargo.com” from any device/pc/whatever without having to think about adding “mobile.” or “m.” before the domain. It also means that as a web developer, I do not have to worry about a different protocol and a completely different way of building things.

“Firefox is the best and most effective platform for getting to that one web.”

Arguably, the application stack that Firefox is built on is the best platform for the web and the brand is second to none.

“We are building firefox for Mobile to spark innovation and excitement.”

Face it folks, innovation in mobile has not taken off as it did with the PC.. yet. There are many great ideas but going to market is really hard. Getting “on deck” or “in catalog” is many times a non-starter. Worse, operators basically have a chokehold on direct-to-consumer. Guessing here, but if these chokeholds were in place during 80s, the PC would not have taken off. (Sure there was “certification” programs that Microsoft had for boxed software, but developers did not have to participate to ship software).

“We need an open web based development platform to do this. We view Firefox as this platform.”

Mitchell, for the win!

My Stuff Search - Use Case

Mitchell also had a great use case of her usage of Firefox on the desktop. She routinely searches for “wheat free”, and the Awesome Bar searches her own history and bookmarks (as apposed to searching the web). Her need when she is mobile, is to access this exact content from her phone. Yeah, that feature must happen.

It was good to see Mitchell talking about Mobile. Our Mobile project is ramping up very quickly and we are making lots of progress. Now is a time to get involved! Please check out http://wiki.mozilla.org/Mobile and see where you can help!

John Slater put together a logo called “Powered by Mozilla”. The idea here is to show your users that the same rendering widget that is used in the Mozilla products — Firefox, and Thunderbird — is also used in your product. Check out this post.

There are bunch of questions about around this effort and getting feedback is crucial to making the successful. Here some the questions at come to my mind:
Administration and review. How does a software developer get this icon. What does the actual process look like? I would hope that this is not very complicated in any way and it also has to respect the nature of many closed source commercial vendors that cannot discuss anything about their product before it is released without a nondisclosure agreement. I would love it if if you could just go to a webpage type in a little bit of information such as who the person or company that is this being granted the use of the logo, agree that it can be revoked, press a button, and then be able to directly download the logos. I’m really hoping that an application review by human is not required. Is a require a review required in order to protect such a trademark?

In the post there is a mention of lineage. I guess it relates to how much source code was used from Mozilla in making this embedded application. For example, if I just used NSPR, could I use this logo? We need to carve out what pieces of the Mozilla stack or required to ship as part of “Powered By Mozilla”. I would argue that you must use most of the application stack — you couldn’t just use the JavaScript library and use this logo in your app.

Furthermore, how fragmented can this embedded application be? Suppose I hated SVG and I thought the web was better off without it. I rolled my own browser and used the entire Mozilla stack, but I disabled as SVG. Would that qualify for this program? I would hope not. We don’t need a fragmented web.

It shouldn’t be this program’s business to verify, review, or think about quality issues in the applications that use the logo. Yes we all want better software, I’m just not sure that this program can or should be used as a stick to get better software. It’s way too subjective, you risk the good ole boy system, and no one has time to QA a stream of products that may be applying to this program.