Building Firefox OS | Hacker News Reader

Building Firefox OS

buildingfirefoxos.com

168 pointsbpierre71 comments

You know, what's great about this new web and these new methods of laying out UI is this: before, content arrived to your browser, and sometimes if you were using a slightly older browser or one which wasn't really tested it'd look somewhat wrong. Useable, but wrong.

Nowadays, it looks like this: http://imgur.com/a/Z7IXu

I'm slightly wary about the future considering these improvements, since we're talking about a page displaying an alternation of a bit of text and an image, a bit of text and an image, ... which I used to believe was something we'd come get right.

I would suggest using a modern browser, it looks fine in FF 18.

I think you missed his point about how content shouldn't turn into a garbled mess of nonsense pixels just because your browser happens to be 3 months old.

True, but undefined behaviour is undefined behaviour, which is probably what's being triggered here. You can't really expect it to always be "slightly messed up but usable."

They're probably using some new APIs and didn't consider degradation at all.

Being 3 months old does not explain nonsense pixels. That's a serious bug in whatever browser the commenter is using - possibly an exploitable one since it got the browser to read from a wrong memory address.

What's the browser?

Hah, I had the same problem using latest Debian Iceweasel browser to look that page. It almost filled up my 32G of Ram and crashed too.

Outdated Google Chrome displayed it perfectly, hell even Lynx did a better job than iceweasel

Looking at the source for the page, the HTML is clean and there is nothing particularly fancy or out of the ordinary about the CSS. It does use semantic HTML (menu, section, nav, etc.), but that shouldn't be a problem.

I'm really curious as to what browser that is, seems like you have triggered some weird bug.

I (sort of) understand the argument, but it does feel kind of out of place since the specific objective of that page is providing guidance on building UI for a specific mobile platform that is build using a modern browser at its core :/

It works perfect on Firefox Nightly. And in fact, they state on the top of the page that you should use Nightly.

> Nowadays, it looks like this: http://imgur.com/a/Z7IXu

What browser is that in? It looks like a weird bug is showing there, potentially an exploitable one since some of that looks like garbage pixel data, like it's reading from invalid memory. That seems to be the issue here, not that web content isn't like the "old days" when it always worked but was sometimes wrong.

Speaking of which, there never were any "old days" that I can remember. Often sites only worked in IE, not just rendered properly but also only worked. You couldn't enter data into forms, or couldn't get them to submit properly. The web wasn't just a little incorrectly rendered in the IE days, it was totally broken on many sites.

> What browser is that in?

Camino, which uses FF 3.6's engine.

is the ff3.6 engine still usable for more than half the web?

As far as I can browse, it's usable for about 99% of the web still. 99% of the one I visit anyway. Although Camino itself does have issues when site javascript locks up and 150-or-so tabs are open.

Fuck Tesla; it's not fair those cars get to run on electricity.

Congratulation on having produced the "complete failure car analogy" of this thread, and thank you for your contribution.

The FF 3.6 engine is very outdated at this point, and there are a lot of known exploits. Unless Camino has been doing their own security work, but I assume not. I'd recommend upgrading.

It is a great time to be a web developer! I have no idea whatsoever if Firefox OS will gain enough of a stronghold to be a viable competitor in the mobile space. I have high hopes, however, that it will spur the technological advancement of HTML-based mobile apps (aka hybrid apps).

Today, hybrid apps are kind of second rate applications. Facebook (on android?) was the most prominent case, but examples abound, where the UX for an hybrid app is leagues away from a native alternative. Better foundation (frameworks/OS support) can change that, and one must applaud the unification of development stacks.

Hybrid apps are those which mix the native languages with Web languages. in Firefox OS and Chrome OS the web is the native environment. There are no hybrid apps, just web apps.

I think it is common understanding that "Hybrid apps" are apps that run on the device, and are written with HTML, CSS and JS and leverage the device’s browser engine - not the browser - to render HTML and process the JavaScript.

PhoneGap/Cordova/appMobi apps are Hybrid apps.

Well, you could make an argument that Native Client is a hybrid approach, but in general I agree with you.

I'm excited about Firefox OS, in concept. However, I'm waiting for something, anything, to show that this isn't going to suffer from large project scope + inadequate resources death. Web controls? Great, but the developers have admitted that they aren't spending that much time on Gonk, etc. because they are UI/Javascript guys and don't want to focus on the underpinnings.

Think about all of the things that Android and iOS don't do that Linux is capable of doing. Firefox OS could be exposing all of that to Javascript, but instead they are focusing completely on the UI and it will always lose to Android and iOS. The power of FF OS/B2G should be the flexibility of allowing HTML5 apps/Javascript to have access to the hardware.

You are mistaken, they definitely are adding *nix capabilities to JavaScript. Go read the B2G or WebAPI mailing list and you'll see they are working on every part of the OS, not just the UI.

The problem that I saw in the API was that it was not open to new types of devices and there was no apparent way to download and install new drivers. That is basic OS stuff if this is ever to be a desktop replacement, and that is why Chromebooks aren't selling like hotcakes.

The WebUSB api has always been on the roadmap, but it got deprioritized for 1.0. It takes a long time to finish all these pieces, so some patience will be needed. :-)

Thanks! Will try to be patient, but it's difficult. :)

Would it allow me to connect a USB keyboard (for example) or a USB mouse to my phone? How about a USB webcam that requires an additional linux driver to be installed? How could that be installed easily by the user? How about connecting a printer via USB? Will I be able to print from my HTML5 app? If not, what will it support? I'd also like to be able to find devices and printers via wifi and print, scan, etc. Will I be able to do that?

I'm not sure of the exact details, but I will try to find out and get back to you.

The ultimate goal is if somebody wants to do something, then it should be possible.

>Think about all of the things that Android and iOS don't do that Linux is capable of doing.

Like what, exactly?

Especially considering that Android is a Linux…

To not only stage FirefoxOS as a potential desktop OS replacement, but 1-up Android and iOS, how about:

* Add support for various types of USB devices. Anything that is supported in Linux via a driver you could install via FF OS would be available to HTML5 apps. Printers, displays, thumbprint scanners, etc.

* Similar support for bluetooth devices, etc. as long as the driver could be coded to work on the device hardware, obviously.

* Gonk should be more dynamic, letting you declare drivers, etc. that you want to use via a path. It's primary purpose would become managing permissions between HTML5 apps and devices. As new devices come out, those could be added easily either by discovery or manual install. When the OS-level drivers are available, it would only be a matter of the HTML5 app declaring what it is able or needs to use and then it asks you (or you configure it) to allow access to the drivers/etc. and then your HTML5 app could use your new bluetooth retina scanner, for example, or control some new helicopter+videocam toy.

The way Firefox OS is now is just a Javascript replacement of Android's interface and it provides a similar level of access to the hardware. That is not enough. I don't want or need a Javascript phone and Javascript gaming device. I want to be able to use the same OS on my desktop as my phone, and if my phone has a USB plug, I should be able to plug one side into my Firefox OS phone and the other side into a label printer and print a label from my HTML5 app if I want. Is that level of integration ever going to be possible?

Yep. The entire goal of this project has always been to fill in all the gaps in web apis like these you suggest.

We have added a whole bunch of things in Gecko and Gonk to support Gaia, which is mainly the front-end. WebAPIs are one example, but we have done much more. Good thing is that this improved the rendering engine of Firefox (that uses the same code base) too.

FF OS should have full access to the capabilities of the device's hardware and anything that it can discover via bluetooth, whatever it is attached to via USB, etc. as long as someone can take an existing Linux driver, recompile it as needed, and then put it in some FF OS repo (official or unofficial) so that it is easily installable from the device itself. Basically, I want mobile Linux as it exists now, but be able to write and run HTML5 apps to access all of the same functionality.

I think if the goal were to make mobile devices as capable as we should all expect but have been conditioned NOT to expect, then that would be an OS that could compete. Otherwise, it is just a slower UI with fewer applications.

Bluetooth is in 1.0, but only headset support and file transfer support because of time constraints. Arbitrary serial connections coming soon.

I just went to a Firefox OS App Day yesterday here in Brazil. I was able to play with a test device and try out some of the WebAPI stuff and I must tell you it is very refreshing to see a system perform so well on such simple phones.

You have access to the hardware but crucial low level stuff requires certified/signed app status.

Check out: https://wiki.mozilla.org/WebAPI

For more information. The cool thing was that my little hosted toy app was installable on the test device, on android with aurora and on my desktop mac also with aurora.

So these aren't necessarily accessible from the device. You need to go to the github for gaia styles and copy them into your third party app.

For instance: http://buildingfirefoxos.com/#progress_activity, you can't just add this stylesheet to your app and get all of the associated styles, but you can go here and get them: https://github.com/mozilla-b2g/gaia/blob/master/shared/style...

You can just include them from your app, those are shared style sheets. Everything in the /shared directory is accessible from third every app. This also includes js helpers.

Also, hi :-)

Hey buddy. How would name resolution work then if your app actually had styles or js in a 'shared' directory? I tried this quickly on Friday but had no luck. Still, I didn't look too much into it. Have you been able to import these into an app of yours?

Not sure, I am merely a C++ dev :-). Surely people in #gaia on IRC will know how this is handled.

Also, yes, it works, see https://github.com/mozilla-b2g/gaia/blob/master/apps/music/i... for instance. This app imports a bunch of code from /shared. Great for code sharing.

But music is a built in Gaia app. I believe these styles to be shared across the "Certified apps" that ship with the phone: https://wiki.mozilla.org/Apps/Security#Certified_application

An example that was a third party app, not certified or made by Mozilla or partners, would be more definitive.

See https://wiki.mozilla.org/Gaia/Shared for the details on how this works. Not having this mechanism would be pretty crappy.

See https://github.com/taratatach/FFDoku/blob/erwan/index.html for a third party app that reuses code from shared/ (a sudoku game).

RE: the first link, it looks like you need a custom profile, therefore a custom build of Gaia, for it to "Probably" work.

FFDoku just copies the assets, as I had originally recommended. That repo contains a 'shared' folder in the root dir, so it's not using the assets shared by Gaia, it's using a literal copy of them. So the same asset winds up on the device twice.

I'll check tomorrow, there is a bunch of gaia folks at the office. You might be right, in which case we need to address that.

Nick, you were absolutely right, and I was wrong. This is a pain, though, but is understandable. I'll ask the right people if we can do something, because I like not copying stuff all over.

Or maybe we can we can do a static analyzer to make the whole process easier.

No worries, we were both unsure. :fabrice verified this today with me.

Is this created by a third party based on own design ideas, or is this by a third party, based on ffOS design guidelines (do they exist already?) or is this published by mozilla?

It looks really great, but those were some questions that came up in my mind when I saw it. It never hurts to have a bit of description on the page.

*edit :

After doing some research, it seems to be a showcase of the Gaia Design Building Blocks page which can be found here: https://wiki.mozilla.org/Gaia/Design/BuildingBlocks

Following the links at the bottom https://github.com/rnowm you'll see the site was developed by someone from Telefonica, a network operator that will feature Firefox OS devices. https://blog.mozilla.org/blog/2012/07/02/firefox-mobile-os/

What's (still) a little mindbending to me about all this, is that you look at that code, code for making native apps for a phone, and it's all just HTML.

It's not for a mock-up or anything. That's what you'd use in a real app. That's lovely.

I see that they're using data-* attributes for a number of these controls- is there associated code for these as well? Markup is great, but a rich control model is more critical for creating great mobile experiences.

Could you clarify what you're asking for?

I like how intuitive and thoughtful animations in Firefox OS. They are also discoverable and consistent, way better than those in Windows 8 and Windows Phone.

What excites me about Firefox OS is how HTML makes it so much easier to programatically lay out all those UI elements. Android's XML layouts are okay with me but look at how there's much less code to be typed in FFOS to achieve the same effect! I guess the same can be said about iOS?

I personally fell in love with qt qml layouts recently, just because having a layout and styling syntax based off something like JSON rather than XML is so much nicer.

But then again, I hate xml. So much useless overhead.

Ubuntu's phone OS will be using QML, fyi.

Blackberry 10 Cascades is also based on QML syntax, though they implemented their own rendering engine instead of the QML implementation in Qt4.

I wonder whether these interfaces have the same lag as Phonegap/Cordova interfaces....

I'm not a big fan of the selectors and date pickers taking up the entire screen. I suspect there will be a lot of hitting "Cancel", reminding yourself what it is you're setting a date for, and then setting the date again.

What's with these bootstrap-like fixed menus on the left? On my laptop, at 1024x768, it does not fit on the screen and does not scroll :-/

If these are going to be the next fashion item, they better be usable.

Strange, the buttons on the examples are click-able when I have NoScript on, but if I turn it off they suddenly become unusable.

WebOS is dead, long live WebOS!

It hasn't been "alive" as a reasonable, retail-ready product for over a year and a half now.

That's still more "alive" than Firefox OS currently is.

Technically the "revolutionary" use of HTML5/JS to create an entire phone's UI/OS/App's is entirely WebOS and it continues to live on.

Little is new, most is old anyways :)

I thought this was about building the source; something I for one have tried and found a point of pain.

FF(O)S, people!

Invest in some cloud build infrastructure (binary blobs excepted; wrap the existing glassechidna client for that), expand your device count, and help people try out your code. No try, no show and tell. No show and tell, no excitement. No excitement, no devs. No devs, no win.

So building from source is pretty trivial. Start here: https://developer.mozilla.org/en-US/docs/Mozilla/Firefox_OS/...

And we do have nightly builds for the desktop emulator: http://ftp.mozilla.org/pub/mozilla.org/b2g/nightly/latest-mo...

"Expanding device count" requires quite the bit of engineering and working with device manufacturers. If you think it's easier than I claim, here's a starting reference and we accept pull requests: https://wiki.mozilla.org/B2G/Porting

EDIT: Also to the above point: https://autonome.wordpress.com/2013/01/15/firefox-os-devices...

For now, the emulator works, third party devs in target markets are getting devices, and a developer phone is shipping in February. And we're having hackathons around the world this week: http://news.softpedia.com/news/With-the-First-Firefox-OS-Pho...

What else can Mozilla do to help create excitement? That's a genuine question, not sarcastic.

I think you need to hire some really kick ass designers and make this stuff pixel perfect.

I'm not sure what fonts are being used but it doesn't look "Apple" polished to me.

The home screen icons are not as good as they could be either.

Maybe a developer portal page linking to the desktop simulator download? This would feel more "official" and "complete" than a wiki page, or grabbing a nightly from FTP.

Android's success was the way it brought in early developers and enthusiasts. The HTC/T-Mobile experiment could just as easily have failed and Android may have failed along with it, but it had developer attention and momentum, as well as an entire 'fanboy' set waiting for it on launch.

The Building FirefoxOS site sounds like it could fill a similar niche as Burning Edge did for the first Firefox (post Firebird) releases, highlight what's new in each nightly release.

I hope we will see more capable hardware and global launch, this is the third platform we need.

EDIT: Sorry, I didn't mean that Building FirefoxOS could fill that niche, only that site similar to Burning Edge could.

Thanks for your reply. I know sometimes it's hard to take a step back and see others' experience, so let me try to show you what I meant about difficulty.

Firstly, I am definitely not the garden-variety user when it comes to mobile devices. I was involved in the launch of the Galaxy Series devices and even spent time at Samsung HQ in Korea for this purpose, and have since run custom ROMs on my phone... so came to this with a relatively strong knowledge, compared to most. If you can't make the process work for me, it's not going to work very well for others.

The build page you mentioned is not the right one for SGS2, which is an extremely popular device, yet in the category 'third tier' supported (ie. effectively ignored).

You should get someone from Samsung to solve this. They have entire skyscrapers devoted to different phone platforms at their headquarters in Suwon, Korea, made vast and record profits this last quarter, and it should not be an issue for them to provide a useful contact for their most popular devices.

Next point. Since the build process itself is IIRC is only supported on Ubuntu or something, I had to download that and set up an entire VM before cloning the code. Second, cloning the code takes years on a third-world connection (literally hours and hours, it's something like a few GB in size).

Third, it wasn't hugely clear that I needed to upgrade from a custom ROM with Ice Cream to a stock ROM with Ice Cream in order to install (in fact, the notion that the existing firmware would be hacked in to the new FFOS image was really not clear at all, and should have been). This resulted in builds failing in non-obvious ways.

Realizing the situation, since I didn't have a stock ROM image just lying around, now insert many hours of downloading different ROM images, trying to install them, getting errors with them. If must have downloaded and tried at least five of these already.

There were also lots of fun errors related to a bug in recent versions of the upload tool, which bricked my phone. Thankfully, because I am somewhat familiar with the Linux boot process and persistent, I did manage to unbrick the phone an eventually get what might be a valid Android ICS ROM going that might be good to install FFOS.

Unfortunately, after spending more than two days on this already, I simply didn't have time to continue prior to my current period of travel.

I am motivated to help, pre-experienced, willing to flash my device and use it on a daily basis even if it's going to reduce functionality vs. Android, but... FF(O)S, people! You can't provide a faster route to an install?

What would have made things easier was a source for something I could 'just flash'. I don't care if it's semi or illegal and it comes from a torrent. I have to go through those routes anyway just to get from Custom ROM -> ICS stock -> FFOS. Give me a script, give me a daily build.

The reality is that with the time pressures to get Firefox OS running and working on the devices intended to be released there has been very little resource to supported other hardware. At some point this may change but it's really up to those interested in that hardware putting the effort in from the looks of things.

Mozilla can't provide an image to flash due to licensing restrictions on binary blobs for the hardware (graphics drivers, video decoders, etc).

Reply on news.ycombinator.com