I recently bought a chromebook, and was surprised at how nice the user experience was. Tasks like opening tabs and loading pages seemed subjectively faster than doing the same on my 4GHz desktop. I also like how my attention was never consumed by irrelevant details like it always is in a desktop os.
In places where GNOME treats the user like an idiot by needlessly hiding useful functionality, Chrome makes specific tasks quick and painless. Surely, the fact that Chrome tries to do less, and benefits from a clean slate helps this, but there is still a very clear winner in this comparison.
I'd use it (chrome os) a lot if it had decent native text editing, and wasn't a botnet.
Hopefully a WM comparable in performance and UI design to Chrome's will be supported in GNU/Linux distros soon.
I think as long as google voice / hangouts is competing for that position, it's unlikely. Skype for linux has historically been kind of a disaster, so I don't expect a chrome app effort from them either.
Unfortunately the recipients I have in mind use skype to talk to their kids & friends. It's non negotiable. My guess is that there are a lot of use cases where 1 or 2 missing apps/features are non negotiable.
IMO the solution is to gradually eliminate barriers to switching from windows. This probably means treating the initial hypothesis that the web is all you need as a starting point and trying to keep chrome's complexity down to Android levels while gradually adding "bloat".
I still think ChromOS could take a big slice of Window's market share, maybe bigger than Apple mac.
That's silly, WebRTC sites are popping up every where. My personal favorite is appear.in, you can have your family chat with you just by them clicking a link. What is it about Skype that makes it the only solution they're willing to use?
Skype has also been a disaster on other platforms. Even the Windows one is full of weird UI layouts and adverts and the Android one is buggy as anything, as is the iOS one on my mum's iPad4. Loss of sync, audio, video, high CPU usage etc. all in there.
Agreed ... well, mostly, I've got one of my machines with an older version that always hangs up when I try to answer a call. But that was an "upgrade", and if I'd tried to upgrade it again after noticing I'd probably be OK.
But the current version for Debian wheezy works great, and I just switched my 80 year old father to Debian with Gnome Classic (he has very few windows open at a time so it's OK) and that version of Skype, and he's quite happy with the result. Heck, he might even be using Chrome instead of his familiar Firefox if Chrome's tabs paid attention to how big fonts should be. Awful oversight for those who's eyesight isn't so good.
Last time I used it was a year ago, and it worked, but I consistently got pixelated, muddy and laggy video compared to both the Windows and OSX versions. And it wasn't driver issues, any other GPU accelerated graphics / video decoding work fine. And the UI was different, along with missing features. The slow release cycle didn't help their case.
Looks like 4.2 is out and might have fixed these, hooray! The UI sure looks nicer now.
Well, it works, but not quite perfectly fine. I have two Linux machines (Fedora and Xubuntu), and both sometimes have Skype crashing when there's an incoming call. There are also random crashes and glitches when Skype is not the only application trying to play sound. I think "disaster" is quite fitting.
I notice that the Android text editor leaves much to be desired too. The wrapping algorithm appears to be really slow, particularly evident in GMail on a tablet if you happen to be writing a reply longer than 2 sentences long (particularly apparent on my Xoom).
> Tasks like opening tabs and loading pages seemed subjectively faster than doing the same on my 4GHz desktop.
That is because your desktop can also do a lot more (and likely is doing a lot in the background.) What is basically a bootable browser, not so much. If you disable unneeded services and drivers in your desktop OS, and have nothing more than a browser open, I think the speed would be comparable.
There's a lot of crazy that goes on in a desktop OS that must be cruft left over from legacy.
I use windows all the time. If you put a bad CD in it you can lock the whole window manager up while it's thrashing the CD to be head or tales of it. Sometimes crash it. As soon as you hit the hardware eject it will free back up.
Reading a CD shouldn't do that. There was the post on HN some months back about the tickrate in windows being set by random applications super low and not being put back up so it uses lots of extra power for no reason.
I just wanted to say the hand wavy "Yeah but X does a lot more Y" as some kind of excuse is all too common.
>There's a lot of crazy that goes on in a desktop OS that must be cruft left over from legacy.
I agree with you but it depends on what you mean by cruft. There aren't really any windows features that actively consume (noticeable) resources and are not useful to someone. Designing feature rich software for a billion different usecases is quite hard. Its reasonably easy to make one-off products that just do one thing well.
> If you put a bad CD in it you can lock the whole window manager up while it's thrashing the CD to be head or tales of it. Sometimes crash it. As soon as you hit the hardware eject it will free back up.
Wow.. are you running Windows 95 ? :) The CD thing was common back then. The only reason this can happen on NT is because of some really buggy driver/hardware combination but even that is highly unlikely. Whats far more likely is you have some kind of buggy explorer plugin that is attempting to generate helpful thumbnails or something while enumerating an unreadable medium. AFAIK Explorer is still single-threaded for the most part for backwards compatibility - which worsens the impact. (Unfortunately a lot of this crap is preinstalled by the OEM) IMO Microsoft made an enormous mistake by allowing third parties to extend their OS in ways that can seriously impact user experience like that. Learn from Apple, MS !
>There was the post on HN some months back about the tickrate in windows being set by random applications super low and not being put back up so it uses lots of extra power for no reason.
Are you talking about timer coalescing? I believe you cant change that setting without a reboot. Or perhaps you mean something else.
Windows Explorer reached its pinnacle in Windows XP when in classic mode and went downhill after that with the crippled search facility (showing two items for each search on Vista - one on the disk and one from the search index it generates painfully slowly), the tabbing to the listview headers (why would you want to tab to a header?), the auto viewing of folders (this MUST be a music directory as it has 1 MP3 file in it so I will show a rating system for ALL files), the giant blue bar that sits at the bottom telling you that you are using a computer with a processor in it (duplicating the System Information window accessible from Windows-Break), the giant useless ribbon on Windows 8, the non-following treeview on the left so that the right and left pane do not correlate to each other (made an option in 7 I think), the top organize bar on 7+ that is not usable using the keyboard, the daft favourite links area that by default occupies 40% of the space on the left, and the flicker-full navigation bar. And another thing: the "uninstall a program" from that silly bar on the top. Explorer is for exploring files and directories, surely not to do EVERYTHING with your computer...?
I always thought the search functionality of windows was horrible till Windows 7 finally managed to deliver a decent implementation of it.
>And another thing: the "uninstall a program" from that silly bar on the top. Explorer is for exploring files and directories, surely not to do EVERYTHING with your computer...?
Well, for about 20 odd years explorer.exe is technically just a 'shell namespace browser' and one of those shell namespaces is the 'explore files and directory' part and the others are the control panel part , uninstall programs part, etc
I believe for windows 95 microsoft originally did not plan to allow third parties to extend the shell namespace because they felt a buggy shell extension could crash the entire OS - which was true for W95. There is an option now to launch these in seperate processes.
I think the issue with the search is that it makes it very difficult to search for text within files, for example. You have to do a search and then modify it afterwards. The same is true in Finder. It's backwards.
What purpose does that giant blue bar have at the bottom of Windows Explorer? It isn't resizable and you can't get rid of it. One bonus is that it has a gradient so isn't ugly, but I can't see the point of it???
The CD thing was really common for me back in the XP era, which also happened to be the plethora of messed up CDs era.
It usually wasn't an OS level event, though, merely explorer.exe, so killing it sufficed if ejecting didn't. (Note: This was when ctrl-alt-del opened task manager directly, so you didn't have to right-click the explorer.exe managed task-bar, which would hang if you did)
Hmm, I don't particularly remember any CD ever crashing explorer, but I was and am very conscious of running a as-vanilla-as-possible install of windows when it comes to drivers, services and explorer plugins/BHOs/Shell Extensions/etc.
> If you put a bad CD in it you can lock the whole window manager up while it's thrashing the CD to be head or tales of it. Sometimes crash it.
I remember visiting a relative's house a few years ago and noticing that they were very fastidious about removing a game CD from the drive when they weren't using it to improve performance. I thought it was a bit silly, as surely no OS would slow down dramatically just because a CD was in the drive.
Of course, this was because I had been using Linux pretty much exclusively for quite a while. The Windows machine (running 7, if I recall) actually was really slow to open file dialogs and other basic things as long as the disc was in the drive. There was nothing wrong with the disc, its just that lots of things synchronously accessed it and weird times.
I made me realize just how much less unpleasant OSes had become since I left that platform.
> If you put a bad CD in it you can lock the whole window manager up
That's because the retry behaviour is quite aggressive - it will keep trying many times, waiting on the driver, until either a read error is returned or a timeout happens (which is somewhere in the region of minutes). The same thing happens with a hard drive that's dying, a slow USB drive, or anything else that takes a long time --- I'm guessing that this long wait is still better than frequent "An error occurred reading/writing. [Cancel] [Retry]" popups.
I borrowed my father's chromebook a couple of months ago and was surprised to see it came with vim, so I found the native text editing not only decent but as good as on any other computer. :) (I think you might need to enable developer mode to use vim, though.)
"I also like how my attention was never consumed by irrelevant details like it always is in a desktop os."
It is odd but I do not find a desktop OS annoying if set up properly. In particular, I felt this wave of relief after reverting to GNOME2 on my Linux VM and particularly this wave of "the OS isn't interfering with me" when moving to Mac OSX. Perhaps it is coming from Windows with the incessant fade-in bubble-shaped popups bottom right near the dreaded notification area that does this? If Windows properly managed this, it wouldn't be so bad.
Such a small difference between the main system desktop layouts but a large difference in the feeling of being constantly flustered from an attention-seeking system. Weird.
(Of course, removal of "distracting" user-interface elements can go to the extreme, as shown in the release notes for Xcode 5 where the toolbar has been refined "to remove distracting elements" or the over-the-top UI culling that has taken hold of GNOME3, eg. the recent changes to gedit as highlighted on the Planet GNOME blog, apparently for the sake of "making it less distracting". Balance is needed!)
Notifications in Windows can be configured, and often come from third-party programs; I use Windows and rarely get any system tray notifications (that's what those popup balloons are called) - the most common one I see is when the state of the network connection changes, and that's definitely useful information for me. I've tried OS X and I can definitely see how some people like it, but I felt like it was too "opaque", like it was hiding a lot of info from me.
> Of course, removal of "distracting" user-interface elements can go to the extreme
IMHO UIs have already gone a bit too far in this direction. The funny thing is that I have a feeling a lot of the same people who think this often useful information is an "unnecessary distraction" are the ones who do not use adblockers and such when viewing webpages, and seemingly have no problem consuming the content with all manner of attention-getting adverts spewed at them.
Very true about being too opaque on Mac OSX sometimes. It forces you to run with an iTerm open all the time to keep your Unix skills sharp, which is a good thing, right?
I prefer falling system tray notifications "big balloons" :-)
Btw, I am not trying to dismiss Windows. I use it daily and love it. But sometimes the inundation of popup balloons is a bit overkill, but that's only when using infested computers...
I played with a c720p last week, I was impressed by the responsiveness of it, I assumed it was a good hardware basis and never thought the software architecture was different.
So they’re replacing GTK+ with something called Aura. What is Aura? “The goal is to produce a new desktop window manager and shell environment”¹. So they’re doing another Unity?
No, it's really not Unity, it's just a custom in-app toolkit. It's a lot closer to Mozilla's XUL layer if anything.
The funny part is though, that they intentionally disable 3D video acceleration, but are happy to use an OpenGL-based toolkit.
The only hope is that they implement a "style engine" ala Firefox so widgets can at least look native-ish rather than Chromium sticking out like a sore thumb.
edit: Oh I read the document you're talking about now. The Shell is specific to ChromeOS. It lets them use the same Aura code in the compositor and the apps, so they get a win there. It's very likely this is to go along with ChromeOS's switch to Wayland or a Wayland-like display protocol at some point in the future.
I still don't get what Aura is, but thought 'Oh, are they creating XUL now?' instead. So far these design documents aren't enough for me to understand what this is _actually_ for.
So I don't have to work with HTML/CSS then? I seem to have a disorder which makes me dislike a separate language for layout/design than the one I'm writing the other code in. And although I do a lot of it and am good at it, I don't like HTML/CSS at all; I have not seen a real world case outside documents and informative sites (which are documents anyway) where I could not build a more robust (and a lot faster) interface in less time in Cocoa with Objective C / C# / C++. No clue what is wrong with people trying to do everything in HTML/CSS. </rant>
Anyway; being lazy; so this is kind of a GUI layer on top of HTML/CSS?
This is nothing like XUL. It's just some abstraction layers written in C++. The hygiene from the Chromium side is that it allows us to share some code with abstractions we already have for the web platform.
Chrome UI on Windows, ChromeOS and soon Linux are comprised of a few layers:
Views (widget toolkit abstraction) - tabs, buttons, textfields, etc.
Aura (window/"native widget" abstraction, event dispatching, what most code at layers above would consider to be the XWindow, except the XWindow is abstracted (and there is only one XWindow per top level browser window))
CC - (Chrome Compositor, also used for web content.)
They're all in native code since being able to drop down to the metal (& grab the app's toplevel XWindow) is often useful when building a desktop UI. HTML/JS/CSS binding to CC is limited to Blink.
OSX users are perceptive enough about the nuances of cocoa that recreating all the UI quirks and what not is a huge undertaking. If I had my sanity intact, I would just stick to cocoa.
I used to be more religious about this sort of thing... but the religion falls apart even on OS X when you realize that even WebKit implements its own copy of the native controls that can appear in web content - so that they can be transformed and animated using CSS.
Feel the same way, however most of the projects we are asked to work on require web UIs. :(
If the browsers used something like XAML, with proper support for components, it would be great.
Maybe web components will eventually get us there, but it is surely a very bumpy road to get to something that only now starts to resemble Smalltalk/Lisp environments.
Why don't they fix the issues with the current UI that aren't GTK+s fault before they go and ditch it and introduce a whole new world of pain?
As far as I can tell, Chromium on Linux has been useless on high DPI screens since its inception. On my 15" laptop at 1920x1080 (just 140dpi) the fonts on browser tabs, and the tabs themselves, are too small and cannot be changed. I simply can't use the browser because of this.
Firefox, which has its own abstraction layer but uses janky old GTK+2 under the hood, does just fine.
I really don't care about extreme OpenGL rendering (laughable, aren't most accelerations disabled already due to driver whitelisting?) or the tiny fraction of the UI that is noticeably GTK+
Great, but let's hope it doesn't start looking crap on regular DPI displays. You know, like those hipster websites that make your eyes bleed if you try to read them on anything other than a Retina display. "Most people I know use MacBooks anyway"
It feels like a premature launch. When I got this from chrome-unstable a couple of weeks ago, it was a shock. Many styling problems, and widgets not working as expected. Some of the stuff that didn't work was pretty niche (e.g. horizontal scroll-wheels), so it's understandable nobody noticed it. But there's even stuff like central UI elements (tabs, the URL box) being rendered in some big, ugly and badly hinted font instead of the system font. Menu items having just insane amounts of whitespace, etc.
There might be good technical reasons for the change, but on a UX level it feels unfortunate :-( I still remember how nice it felt to switch from Firefox to the initial Chrome Linux alphas. It might have been missing major functionality, but at least it looked and worked the way I expected a Linux desktop app to work instead of having the UI be just a little bit off in a dozen places. I'm sure they'll get it to a 99% state soon enough, but getting that last 1% looks like it could be a challenge.
I'm not treating it as it's stable, but it's what's going to be released as stable in a fairly short while given the Chrome release cadence. I was hoping that they were going to let this bake for a bit more before making it the default. This post makes it clear that they aren't going to do that.
They don't seem to do a straight unstable -> stable path. I'm not completely sure, but I'm pretty sure I've seen features stay in unstable while stable releases keep happening.
I'm totally fine using unstable, thanks. The point is that this particular feature is going to be enabled by default in Chrome 35 on Linux, as per the submitted article. That's not going to be 6 months from now. What happens with unstable features in the general case isn't particularly relevant when we know what the plan is with this specific case.
I found the last 2 comments in that thread ironic. One guy's asking why Aura is a window manager, and 'mike' says Aura is not a window manager, Ash is the window manager, but then proceeds to supply a link which has 'aura-window-manager' in the url! His link doesn't even mention the word 'Ash'.
They already made their own thing for Chrome OS rather than use GTK/Gnome or Qt/KDE there.
I find it sad that they felt the need to make their own UI engine vs being able to contribute the changes they feel they need to an existing project. I'm sure Google engineers will tell you that they need to move faster than they could working within the community of GTK or Qt. Plus they don't want Chrome OS to suck because they can't get features merged by the community. They are valid points. The resulting fragmentation just sucks.
I assume Aura is open source. Given GTK's dropping adoption rate and a lot of people unhappy with the recent direction (design or lack of listening to current user base) of Gnome how long until someone writes a general purpose (not specific to Chrome OS) window manager/desktop environment on top of it?
I don't see that ever happening. Google does a lot of reinventing the wheel with their projects and most of it is not built in a way that is easily useable in downstream projects, and they probably won't accept patches anyways. Chromium OS, for example, could have been forked and used in a lot of cool ways but that hasn't happened yet and I doubt it will.
Huge GUI frameworks are overrated. They force you into their way of doing things. Then something newer and shinier supersedes them. If I had Chromium's developer power on my own projects I'd probably create custom GUI tools too.
Yeah, given that they have most of the code already done for drawing the browser screen(text, buttons, icons...), it probably makes sense for them to do so.
With Qt they will depend on a different company with different interest in mind.
I don't think they are "making their own thing". From the post, it seems like this is already what they use on the other platforms.
From what I gather, they are switching the Linux builds to use it as well so they don't have to duplicate development effort. I think that makes sense. It sounds like it might be rough and they seem to be aware of it; hence, the urgent call for testers.
Google knows that the Open Source development process while nice and warm and fuzzy, can get in your way when you don't own the entire development of the project - especially if this is a product that can impact your revenue. Others can arbitrarily (sometimes with good reason) block your feature pushes and slow down your product releases. They did the right thing forking webkit, keeping Android/Chromium in-house, and are doing the right thing here. - IMHO.
GTK+ started making less sense for serious desktop development when Qt went LGPL in January 2009 and then went on to become highly modular in Qt5.
The only 'advantage' GTK+ really has over Qt is that it's pure C, so there a lot more bindings for all the languages you're probably not going to use to write significant desktop software anyway.
I cannot agree more. GTK+ still exists because that's one of the few pure-C toolkits left around.
Qt, wx* and Fltk (which are arguably the other most popular widgets in order of popularity) are all C++. Not to mention that wxWindows on linux actually uses GTK+ for rendering, and it uses a terrible event table mechanism.
This isn't really leaving GTK+ because of GTK+, this is recognising that using three different toolkits for different OSs is hard. ChromeOS would also be a huge reason.
Weird isn't it? First, everyone wanted it to look flat. Then Win3.11 had mild bevels. Then Win95 had more bevels and 3d look panels (sometimes recessed, which just involves drawing the two lines the other side of the panel...). And then things went really squidgy-looking with GNOME, and then it has gone flat again.
I am not sure what people want. At least buttons look like buttons (unlike the way iOS and other mobiles OSes are going). Is it a button? Is it a plane? No, it's text! You end up bashing your screen as it isn't obvious what to press. My mum is really confused.
You can change many aspects of the look of GTK+ with themes. You can probably even make it look nice and flat, in keeping with the latest UI fads.
Personally, I think XFCE looks nice out of the box (in Xubuntu, at least). And what's a better alternative? Unity and GNOME are both way too dumbed down for me, and KDE is both ugly (IMHO) and and quite buggy.
As someone who has only recently jumped into linux desktop development, what's the alternative? I mean, elementary have a nice fork but it's still GTK (albeit wrapped in Vala which takes the pain away), and don't get me started on KDE.
Qt/GTK+/FLTK, depending on the language or footprint. Of those, Qt/FLTK offer very easy multi-platform development.
GTK+ was always incredibly poor in anything but linux, and with GTK3 the footprint/performance is worse than Qt.
wxWindows has a very ugly event table/dispatch mechanism, which somewhat reminds me of the old win32 api. On linux it even uses GTK+ for rendering. It used to make sense before Qt4 came, but nowdays I wouldn't use it.
FLTK or wxWidgets. wxWidgets on Linux is actually wxGTK though, but it's pretty painless. Or you could use Qt, which doesn't really look native on other platforms.
Hopefully this finally means retina display (HiDPI) support for Windows builds - currently Chrome is almost unusable on all high-res displays on Windows.
In the olden days, only the X11 version was open source, but it was made available under the GPL on all supported platforms with Qt 4 and under the LGPL with Qt 4.5.
In places where GNOME treats the user like an idiot by needlessly hiding useful functionality, Chrome makes specific tasks quick and painless. Surely, the fact that Chrome tries to do less, and benefits from a clean slate helps this, but there is still a very clear winner in this comparison.
I'd use it (chrome os) a lot if it had decent native text editing, and wasn't a botnet.
Hopefully a WM comparable in performance and UI design to Chrome's will be supported in GNU/Linux distros soon.