Post a Comment
Actually I'd say this is a symptom of a MUCH uglier disease which is thus: The ONLY reason we are seeing HTML V5 pushed is because of Cupertino using their muscle and frankly its just not ready yet.
Apple did NOT pick HTML V5 out of the goodness of their hearts, they did it to kill the market for Flash games. But HTML V5 isn't ready, it can't even do a third of what Flash did and what it does do it does poorly because its just not RTM quality yet. take ANY video, any res, and put HTML V5 VS Flash at the same resolution, HTML V5 H.264 will suck up cycles and memory like a drunk hitting a free mini-bar, its just awful.
So the bigger problem is how to fix HTML V5 without Apple ramming it down our throats before its ready. As Moz is finding out you have to have seriously beefy hardware just to get the stupid thing to run and with "thin and light" being the order of the day if everything goes to HTML V5 i hope everyone likes battery life measured in minutes because it ain't gonna be pretty.
The Alcatel one seems very alpha, but the ZTE and Geeksphone ones look OK for the price:
http://www.engadget.com/2013/02/24/zte-open-hands-on/
These things are build to be maximum cheap, 256mb RAM, single core, 512MB storage + SDcard, no patent payments to Apple or Microsoft AFAICS etc
And if Mozilla keeps its promise you will be able to buy apps from anywhere and the phones can be rooted. So no lockin and carriers become dumb pipes.
Does not sound too bad to me (I won't buy one though)
Edited 2013-02-27 23:15 UTC
The only reasons patents wouldn't be an issue is because it isn't popular enough to be in anyone's sights. If it ever does become a concern to them, you can be sure they will be finding appropriate patents that they can claim are being infringed.
If it uses an SD card, it is already in a patent minefield, since complying with the SD standard apparently requires using the FAT32 and exFAT filesystems.
So then, I take it, they're not going to support fat32-formatted sd cards on the grounds of not paying for Microsoft patents? That'll really make the Windows users happy.
You haven't used firefox in a while now, have you?
It's far faster than chromium for me, and uses WAY less RAM.
Chromium with a couple of tabs open will quickly shoot up over a gig of RAM used in total; it's ridiculous, and the process separation idea is totally ludicrous.
It's far faster than chromium for me,
True, this is on GNU/Linux. After all of my addons were installed, it still felt more responsive even when I used to use windows, though (chrome needed much more addons due to lacking functionality).
Well, considering that the RAM problem seems to be related to it, and that with a combination of good session management and load-on-demand after a restart, the issue of pages crashing the browser is alleviated, it seems silly.
Aaron Seigo reposted something on his blog the other day about how it was proving a limitation for web app developers, too.
Edited 2013-02-28 10:49 UTC
Well, considering that the RAM problem seems to be related to it, and that with a combination of good session management and load-on-demand after a restart, the issue of pages crashing the browser is alleviated, it seems silly. "
Silly? Last I've heard they have revived the Electrolysis project for the FirefoxOS ;-)
1) the link he posted, is broken now so there is no way to see if it is a good analysis or not.
2) the reason he posted this is because many criticised KDE's design where one malfunctionning plasma applet can bring down the whole desktop.
3) he 'forget' that Chrome can also be configured to have everything running in one process with a command line option..
I'm not advocating having each plasma applet running in its own process by default, but that would be handy to debug an issue with a Plasma applet, no?
Edited 2013-03-04 14:57 UTC
http://imgur.com/dH3GO8b
You were saying? (I have 6 chrome Windows Open).
Also process isolation is a good idea, because when one tab locks up/crashes I just have to reload the tab. I am sure there are a host of other reason why it is a good idea.
Edited 2013-02-28 15:14 UTC
You were saying? (I have 6 chrome Windows Open).
Also process isolation is a good idea, because when one tab locks up/crashes I just have to reload the tab. I am sure there are a host of other reason why it is a good idea.
You failed to add up all the chrome processes or give any details about how you tested this. That's kind of the point of "process-separation", there are multiple processes.
I've never had that problem; it works fine even for my parter, who for some reason refuses to close tabs, and normally has over 200 open on a windows box.
Problem sites *will* crash the browser, but you just don't open that tab when you reload the browser - they load on demand, so when the browser restarts, only one tab will actually load.
Differing experiences and all, but this is with only two tabs open, and every extra tab increases chromium's usage by a lot more than it does firefox's - it's pretty much as close as they ever come in usage for me.
http://img541.imageshack.us/img541/6753/rams.png
I use modern Firefox quite heavily on Windows 7 at work, and on Ubuntu 12.04 at home. Older Firefox versions suffered horribly from responsiveness and runaway memory problems, and I switched to Chromium at home to avoid it for quite some time. But I recently switched back to Firefox, and the problems seem to have been completely resolved.
Understand that I still like Chrome, and didn't switch away because it had any serious problems, I just find Firefox now works better for me.
(When I say "quite heavily", I mean that I typically have 30-40 tabs open in 3 browser windows on my home machine, and perhaps 10 in a single window at work.)
All that said, I'm about 3,217x more excited about Ubuntu Touch than FirefoxOS, but I'll definitely give them both a try.
You were saying? (I have 6 chrome Windows Open).
Also process isolation is a good idea, because when one tab locks up/crashes I just have to reload the tab. I am sure there are a host of other reason why it is a good idea.
Post your about:memory
I see the same memory gobbling behavior on firefox under both linux and windows.
As far as I can tell it's not related to what I'm doing, just a slow continuous leak of some sort that persists even after I close all but one tab.
about:memory doesn't look very meaningful to me, but right now with nothing but osnews open I see js-non-window at 80MB and js-main-runtime at 90MB. Everything else is much lower in comparison so a pure guess is that's it's a JS related leak. But for all I know it might be one of the plugins I've installed.
On my linux box it usually takes a week or two before firefox starts to crawl at a few gigs of ram. This happens regularly because the system is always on. On the windows side the computer is restarted every day, so although the memory usage climbs, it never becomes unusable.
I haven't tried chrome for comparison.
As far as I can tell it's not related to what I'm doing, just a slow continuous leak of some sort that persists even after I close all but one tab.
about:memory doesn't look very meaningful to me, but right now with nothing but osnews open I see js-non-window at 80MB and js-main-runtime at 90MB. Everything else is much lower in comparison so a pure guess is that's it's a JS related leak. But for all I know it might be one of the plugins I've installed.
On my linux box it usually takes a week or two before firefox starts to crawl at a few gigs of ram. This happens regularly because the system is always on. On the windows side the computer is restarted every day, so although the memory usage climbs, it never becomes unusable.
I haven't tried chrome for comparison.
You can post your about:memory after you feel Firefox has slowed down after 2 weeks. Maybe in a bug to Mozilla or in the forums. Look up project snappy because they would be really interested.
Fergy,
As a dev myself I'd feel it necessary to take out all the variables by turning off all the plugins and confirming the leak still exists. I'm not thrilled at running without adblock/ghostery/firebug for a week.
I'd be kind of surprised if they aren't already aware of the problem, so it's probably a matter of finding a similar bug and saying "me too".
Guess I knew this already so it wasn't surprised to me atleast.
My guess is, this is about unlocking the Apple and Google stronghold and they could really use a partner for that.
As many carriers already pay for the phone anyway and Mozilla has a relationship with them and the customer has a relationship for the payment system with them, so might as well be them.
I can choose my carrier and I can choose the appstore I use. I can even install from a website directly.
Is that not open enough ?
EDIT: now that I have put in some more thought, Mozilla is doing this for the developers.
The idea is that Apple gets a lot of money from the developers, like 30% or even more.
My guess is, they want to lower this number.
Edited 2013-02-28 00:01 UTC
The idea is that Apple gets a lot of money from the developers, like 30% or even more.
My guess is, they want to lower this number.
Not only that but give them a lot more freedom on how they distribute their applications. Since a Firefox OS app is just a web page you can package it and put it on the existing marketplace, host it on your own server both as a packaged app or directly as a web-page, etc... One of the major advantages of this is that if your app is free but you want to offer paid services from it you also can and that's something that Apple has basically barred people from doing as they want all payments to go through their channels.
Well I'm very surprised about 3rd world on any networking — from "carriers" to capped <100 megabit internet with some "six strike program": http://arstechnica.com/tech-policy/2013/02/heres-what-an-actual-six...
Oh, this is America? Sounded like Nicaragua.
Oh you americans.. I remember you labeled Europe as "banana republic" when we declined some retarded software patents.
Hilarious country. Atleast I hope this is not true: http://techrony.com/wp-content/uploads/2013/02/Human-Genome-Project...
Edited 2013-02-28 09:15 UTC
Mozilla working with carriers in an open and transparent way is great news imo.
Sony will probably have better hardware: http://developer.sonymobile.com/2013/02/27/experimental-firefox-os-...
Edited 2013-02-28 04:17 UTC
Could we please stop saying "you are not the customer, X is"? or "you are not the customer, you are the product"? The latter has already been debunked*, and the former... As if it was incompatible? Both you and the carrier are the customers.
*http://www.osnews.com/story/26626/_I_m_not_the_product_but_I_play_o...
Disclaimer: I work on the Firefox OS project so I'm biased towards it.
I wanted to chime in on the first impression many had of the operating system. The phones we're demonstrating Firefox OS on are incredibly underpowered, they've got specs comparable to the lowest-end Android phones. There's a reason for it: they're intended to be launched in emerging markets at very low prices. That's part of the project's goals, to provide a wide range of people currently limited to feature phones with a smartphone-like experience with the added benefit of optimal access to the web.
Yet there's a couple of performance aspects that I'd like to underline. If you'll have the chance of trying the handsets yourself you'll notice that their performance is not bad at all, with most apps capable of a very smooth experience (55+ FPS during transitions/animation is not uncommon). More importantly it's not bad at all when compared to similar Android handsets and it's nothing short of amazing when you realize that's just HTML, CSS and JavaScript code running on slow, single-core, single-issue ARM processors.
In addition to this the OS in general tends to use significantly less memory hungry than Android for comparable tasks, that's an aspect we were very focused on.
Just to show that I'm not making this up check Andy Rubin's comment on this:
http://allthingsd.com/20130226/googles-andy-rubin-on-firefox-os-in-...
From the article:
“There are places where Android can’t go,” he said, referring to memory and other hardware requirements.
Yes but Java and C# have better defined datastructures. One of the problems with compiling JavaScript is that the central datastructure is essentially a hash-table, which makes it very hard to convert to efficient machine code, though there are several tricks to try anyway and then go back to interpretation if it fails.
It's called Type Inference and has been part of Firefox since the start of 2012:
https://blog.mozilla.org/blog/2011/12/20/major-javascript-enhancemen...
Its a neat trick, but how far will it get you? How long before you run out of obvious optimizations? How long before you run into the brick wall that is the language itself?
It just doesn't make sense. JS will always be worse off because it performs a lot at run time, that is done at compile time (with much more diligence) in other languages.
There simply isn't a comparison between the performance of a JS JIT compiler and what a good optimizing compiler or even C#'s JIT can do.
It is true that JS has geniuses working behind it (esp. the V8 team) but JS as it stands now isn't ready, and I'm not sure it'll ever even become good enough.
Javascript is now about twice as slow as C in most cases. Where other scripting languages like Perl are more than a 100-times slower than C.
Where something like Java might be 1.3 or in certain cases faster than C (a JIT compiler can optimise stuff at runtime which normally a C-compiler would not build that into the compiled binary).
Javascript is basically the fastest scripting language right now. Faster than Lua for example, but Lua can still be faster because it is easier to exchange data with existing C-code. Lua is also a lot more memory efficient.
So yes, Javascript never was the best choice, but when did you ever see the best technology win ? Take VHS Betamax as an example. The "problem" is, Javascript is widespread and "good enough".
Also supposedly FirefoxOS uses less memory than Android because there are a whole lot less layers involved. So FirefoxOS might actually win from Android "native" apps on a lot of devices.
I read somewhere a few years ago that in servers memory usage as much power as the CPU. CPU's have become much more efficient. So that would mean Javascript could have similar power demands than Android/Java.
Time to pay more attention to compiler design classes.
Languages != Implementation.
There are native code compilers for C# and Java available.
Even on Windows you can compile C# directly to native code by making use of NGEN.
Are you aware that .NET Windows Phone 8 applications are actually compiled to native code?
Or that there are Java native code compilers for iOS?
And the C# JIT has over a decade of work going into it, and Microsoft's extensive language and compiler experience from even longer than that.
JS doesn't do even a fraction of the optimization even a clumsy C# JIT does. Start doing install-time native compilation of C# and the story gets even better.
It depends a lot on what you're doing, for applications with a heavy computational factor it most certainly won't (though matching Java/C# is not impossible in JavaScript). For applications leveraging the runtime more it can be every bit as fast as a native application from the user perspective.
It is improving and some of the latest optimization frameworks being introduced @Mozilla are specifically geared towards actual web apps rather than regular web pages (the IonMonkey JIT for example).
It depends a lot on what you're doing, for applications with a heavy computational factor it most certainly won't (though matching Java/C# is not impossible in JavaScript). For applications leveraging the runtime more it can be every bit as fast as a native application from the user perspective.
You just don't know enough about the executing code to do meaningful code optimization, and are running under time constraints as the code gen is happening in real time. This is in contrast to the case with C#, especially as moondevil mentioned, where compilation is done offload and gratuitous optimizations can be applied.
It is improving and some of the latest optimization frameworks being introduced @Mozilla are specifically geared towards actual web apps rather than regular web pages (the IonMonkey JIT for example).
Mozilla being on its Nth JavaScript engine doesn't really impress me as much as it once did.
Just take a look at https://developer.mozilla.org/en-US/demos/detail/bananabread and then please comment on whether you still think that JS JITs can't do "meaningful optimizations".
Can you do me a massive favour and communicate to the other Firefox guys that the desktop version is still a pile of shite.
The inspector tools are shit unless you use Firebug, which is third party so it probably leaks fuck loads of memory everywhere.
When are you guys going to get you are spreading yourselves too thinly.
People tend to have this utopia idea that carriers can be circumvented, but as long as they have anything to say that is very hard thing to do.
Only on countries were people tend to prefer pre-pay to contracts this is somehow possible, but people need to be able to buy the handsets, which tend to be quite expensive nowadays.
So if the choice is between a carrier specific firmware or very expensive standard firmware, many people will pick the carrier one.
Given my experience in telecommunications I can assure the situation will never change, regardless of what a few geeks might wish for.
Get prepared for the myriad of discussions about FirefoxOS being fractured and updates only available in certain devices.
Only on countries were people tend to prefer pre-pay to contracts this is somehow possible, but people need to be able to buy the handsets, which tend to be quite expensive nowadays.
So if the choice is between a carrier specific firmware or very expensive standard firmware, many people will pick the carrier one.
Don't assume that your experiences cover even most of the world. Here in Finland, for example, carriers generally sell their phones completely unlocked even if you sign for a 2-year contract, you are also given the possibility of just buying the phone outright and going for a pay-as-you-go contract if you so choose, and I'm not aware of any operator bothering to make custom firmwares.
My experience around the world is based on my former employer right there in Espoo at Karaportti campus, from 2004 to 2008. "
I just have to wonder how you've gotten such experience, then, because what I just said in my previous comment hasn't changed in almost a decade now. TeliaSonera is the biggest, the most-Americanized carrier here and even they sell phones unlocked, with or without a contract of your own choosing.
Yes, it is true some operators in some countries do offer that possibility, but in most countries you end up paying more for 100% carrier free phones and many people avoid it.
Just to clarify a bit my previous comment, although I was going to Espoo on a regular basis I was actually located on Düsseldorf site.
But OTOH most of the 5+ billion mobile subscribers own their phones upfront and are on prepaid, IIRC (from some Ericsson publication)
I used Firefox OS on one of the underpowered devices and on faster devices.
I also used Android on both.
On the underpowered device, Firefox OS is noticeably faster for many operations than Android, in particular, web browsing.
On fast devices, Firefox OS and Android look very similar in execution speed.
Another thing: Firefox OS applications also work in Android. You can install them via Nightly. That's pretty cool.
Personally, all I'd be waiting for is actually good apps.
I have been using Grooveshark html5 app which is excellent (and has no non-HTML counterpart that is as good), but there are still many other apps missing.




