OSNews

Cool. Microsoft always seems to come out with some new API every 5+ years, but the way you know they're serious about something is when Internet Explorer, MS Office, and/or Visual Studio gets rewritten with the new APIs. With Office, it looks like they're on their way.

I haven't finished watching the Build keynote yet (I'm about halfway through), but I wonder just how robust this WinRT API is. With .NET and C#, I found that it is necessary to use pinvoke/Win32 for quite a few things such as clipboard monitoring and MIDI. Hopefully, we won't have a situation where Metro developers are having to 'dip into' the .NET or Win32 APIs to get work done.

As for Win32 itself, as many corporations are still stuck on IE6, you know Win32 is going to be around for at LEAST the next 10 years. Who knows about .NET.

One other thought - when they were showing off that you could have legacy apps in the Windows Store, the one they were showing (can't remember which), it said 'download from developers website), so I wonder if Metro apps will be the only ones you can actually download from the store?

It was Quicken or QuickBooks.
Sinofsky said the store would act as a listing service for non-Metro apps.

From the info from people who've dug into the Milestone builds, WinRT basically offers unified access to the system whether from .NET, Javascript, or unmanaged code. It'll be nice to see how much is surfaced at this point -- also nice to see if any PowerShell enhancements are included in this build (and how WinRT helps make it easier to interact with the system from PowerShell).

I'm sure the details will be on MSDN shortly ( http://dev.windows.com/ ) but until then, this article may answer some questions (of course, you can also download the SDK from dev.windows.com.

http://arstechnica.com/microsoft/news/2011/06/windows-8-for-softwar...

So they're going to host Metro apps, but simply link to non-Metro apps? Well, THAT certainly won't confuse end users

:-) APIs are up

http://msdn.microsoft.com/en-us/library/windows/apps/br211369

So when they say Metro can be programmed in C++, they mean managed C++, not regular C++.

That is a great disappointment for us that hate Microsoft APIs with a passion.

I've gone through the API, and it's horribly complex. Most UI components, for example, inherit from more than 5 interfaces, and have tens of properties and methods, most with cryptic names.

The new style Metro UI is a real anathema for us real time application programmers. It means that we can no longer write our code with Qt, since the Metro UI means the code must be written with .NET. If our clients request Metro UI applications, then we will have to stop supporting Linux and other Unix-based OSes, because the Metro UI is exclusive to Microsoft from all sides (libraries and programming languages).

Microsoft has done it again. Instead of embracing standards, they created yet another vendor lock-in solution.

^FALSE.

Please watch this session, it explains a lot. A lot.

http://channel9.msdn.com/Events/BUILD/BUILD2011/BPS-1005

It's XAML + *NATIVE* C++ (compiles straight into x86).

(Also, according to presenter: "We have reimplemented XAML in native code.")

Edited 2011-09-14 15:38 UTC

Managed C++ still compiles straight into x86.

Based on the examples given for C++, it looks to be a mangled manged C++ with .NET integrated. See "Object" - the based class: http://msdn.microsoft.com/en-us/library/windows/apps/system.object

It directly invokes .NET in the description. Perhaps that is just a fluke, but goes to show where it's going.

EDIT: The C++ side also seems to require the /clr option as the C++ examples use the 'sealed' keywords, (see http://msdn.microsoft.com/en-us/library/windows/apps/windows.ui.cor...) which means its Managed C++ at best, not Standard C++ but Microsoft's hacked C++ instead.

Edited 2011-09-14 17:55 UTC

No, wrong. Standard C++.

Those sort of organisations wouldn't upgrade to Win8 let alone Win9 (or whatever follows on) anyway.

While we're on the topic though:

I'm I the only person that sees this whole drive towards offline HTML apps as completely retarded?

We shouldn't be dumbing down application development further for people who can't be bothered to learn a proper development framework.

Maybe I'm just oldskool, but we should be encouraging the use non-crappy pseudo-languages for doing real development work.

Edited 2011-09-14 12:19 UTC

On one hand, I agree with you. On the other hand, I think it's time to put C++ out to pasture and find (or invent) something else to write native code with.

Laurence,

"I'm I the only person that sees this whole drive towards offline HTML apps as completely retarded?"

No you are not. We ought to be moving to technologies which enable fat apps to be portable and provisioned/migrated transparently. Instead the trend is going to thin apps, which can't make good use of local resources and are dependent on third parties with more modes of failure and higher latency. This downgradeds the user experience.

Some might argue that HTML/JS can be overhauled to fix it's shortcomings, but it's already in such sloppy shape that web developers are cursing it every day. It's just extremely poor at a lot of things. Take vector graphics...terrible support for that, now we're dependent upon the server to render simple shapes. Want to add audio feedback for something, can't do it reliably without a plugin. Want the webapp to inherit the look and feel of a native application, forget it. Making things work in all browsers is a major undertaking, even for a site with a simple layout. One of the most common web layouts is columnar, which is exceedingly difficult to get right using CSS, often times resulting in hanging text syndrome.

HTML was cool at document markup, but as a platform to replace desktop applications, it sucks.

Well, there is Java.

I am actually impressed by how fast Java 7 can run Minecraft. It seems to be a good improvement over Java 6.

It is still a bit ridiculous that I have to give it a full 1 GB RAM to avoid stuttering, but it does run very well.

Thank You!

Try being more imaginative.

You replied to the wrong person. You must of meant to reply to Microsoft/article.

This move to a new UI isn't anything new. Anyone who has been around a while probably experienced a sense of deja vu when Microsoft announced that Metro will be the new default UI in Windows 8.

Remember back in the mid-90s when Microsoft ignored the Internet until it started getting ubiquitous and then panicked and tried to make everything look and work like a browser?

Fast forward 15 years and Microsoft is way behind iOS and Android in the smartphone market, hence they panic again and overreact and try to make everything look like a smartphone, whether it makes any sense or not.

That's not how I remember things.

Windows didn't look look and work like a web-browser until the birth of IE4, which came out a couple of years after Win95 had.

Windows 95 before IE4 still felt very "offline" - like a stand alone computer. However it did still have some core APIs there already and some stuff was already subtly exposed to the users (eg run and file open dialogs could accept 'http://' URI's).

So my problem with your post isn't the "over-reaction" of the IE4/Win98 layout (which actually I did like). It was that you stated MS "panicked when the started getting ubiquitous". From what I saw of the ground work in Win95, Microsoft were always heading that way. It wasn't a knee-jerk reaction, it was a staged strategy as the jump from 3.1->95 was massive enough already.

I'm sure as early back as the early 90s I had read an interview with Bill Gates where he stated he wanted to move towards web-driven software which was licensed rather than bought. I can't for the life of my find the source, so who knows if I made it u or not.

Tell that to Google and Apple. Both of whom run phone OSes on their tablets.

You do know a desktop computer is different from a tablet, right? Here's some links for education

Desktop: http://en.wikipedia.org/wiki/Desktop_computer
Tablet: http://en.wikipedia.org/wiki/Tablet_personal_computer

There are a lot of apps that create multiple windows. Think about opening up a message in Outlook or a conversation window in an IM client. I suspect Microsoft concluded it was better to go with a solution that would behave predictably for all legacy apps, even if it wasn't optimal in some cases.

Then you can make a full screen wrapper for those apps, that just shows that app's windows. Popups are also a bit of an issue, but you can host them inside the same full screen canvas of the parent app.

I realize it might be a bit difficult, and has the possibility of breaking some things, but it allows you to use all of your apps within the same shell and with the same usage model. I'd actually find it to be pretty usable and conceptually consistent.

Alternatively, just entirely separate the legacy and the new. Make legacy it's own little Win7-like bubble, with it's own start menu where apps and settings affect only that environment.

Those apps are written under the assumption that they're running on the "classic" desktop. For example, maybe there's an app that relies on data being drag-dropped from other apps onto it before it can actually do anything (that's just an off the top of my head example of an app that would require the classic desktop).

I think it's better to have a clean separation between the two environments then try to guess what a classic desktop app might need and try to simulate it in the metro environment.

So some legacy apps break or don't work, or need a new release to fix issues.

On small screens, like palmtops, netbooks, tablets, etc (really anything under 1366x768 and 13"), the app menu metaphor doesn't work. Even with a mouse.

Not sure how well a combined 'start screen' and widget display thingy will work on a desktop. But icon-based 'start screens' like KDE's plasma-netbook work extremely well, where the 'desktop' IS the app menu.

GNOME-Shell, I believe, also does something similar, although not nearly as well.

Granted, for 20"+ screens with giant resolutions and great DPI, the app menu still works. I still use plasma-desktop on my 19" disolays, but plasma-netbook everywhere else (HTPC, laptop, netbook). Wouldn't dream of going back to an app menu on them.

I'm sorry I wasn't clear. I'm not talking about Metro or small screens.

I'm talking about the Windows 8 Desktop, where the Start Menu has been replaced with a list of 4 items: "Settings, Devices, Share, Search." It even uses the new Google art style: "White Text On Black Background." In the business we have a name for this: "Fucking Retarded."

With Windows 8, it looks like Microsoft is trying to commit financial suicide. Especially since most of their income comes from businesses.

People said the same about gnome shell and I agree with them. why suddenly change a direction that was working?

I'm waiting for the ultimate in portable convergence:
* a smartphone that slides into the back of a screen to create a "tablet" (where the "tablet" is just a battery and a screen)
* the "tablet" screen includes a port to attach to a keyboard (where the keyboard includes a battery and maybe some extra CPU/GPU/RAM?)
* with an option to slide the phone into the front of the keyboard to use the phone screen as the touchpad

Basically, the phone is the heart of the laptop/tablet, with all the storage, CPU, GPU, RAM, etc. And the add-ons provide extra battery power and possibly CPU/GPU/RAM.

But the way the Motorola Atrix works is just dumb. The touchscreen on the phone is wasted when docked, and the OS on the phone is not used, etc.

Basically, take the eeePC Transformer and add a smartphone dock in both the keyboard and the screen.

Unlikely IMHO. Your vision reminded me the prophecies about what was supposed to happen with home computers in the 80s, using the then-widespread computer technology & UIs; one central home computer which can control pretty much everything else.

It didn't end up that way of course. It turned out that the "computer" / microprocessor part of the equation is really no big deal and almost the only one which dramatically drops in cost over time - now, most of the devices have their own computer integrated, typically at least comparable or very much surpassing the microcomputers of the 70s/80s era. Inexpensive connections between them were a larger issue. Or even, really, giving this pursuit some meaningful purpose.

Your "tablet" part would be probably pretty quickly at least as expensive as full tablets, while being less handy and limited in form factor by the need to have a standard slot for (similarly limited) standard smartphone. SoC is not such a big part of the costs ("tablet" would need quite a bit of electronic junk anyway, probably most easily implemented by... some standard SoC); but you would add a need (and cost) for a reliable, high speed connector.

IMHO only such keyboard (just a simple keyboard, NFC or BT, & battery option) maybe has some future... maybe with a holder for smartphone (but a relatively independent one, wireless synch with & control of the rest)

Great comments. I'm not sure that a tablet will ever really replace a phone -- since the phone has some obvious advantages with form factor that allow you to shove it in your pocket easily. That is, unless you wear "mom jeans" with huge front pockets. LOL. ;-p

Actually I think it's a good news for them.
It means MS is ultimately going to freeze developement of win32. That will allow Wine/Reactos to finally catch up become usefull in real scenarios and occupy a nice niche of providing legacy windows compatibility forever.

ReactOs developer should now pledge some VC to get his milion euro to get "ReactOs 99%" ready to demonstrate on the timeframe of Win8 premiere to all disgrunted corporate users of Windows. A "window" of opportunity indeed .

Think of them as FreeDos next generation. It would be ironic if in future Linux became a primary desination for legacy windows apps.

Edited 2011-09-14 10:12 UTC

Interesting analogy, and I think you are right - unless this transition to a new API fails entirely - in which case ReactOS could be like AMD64 to Intel's Itanium architecture.

There are ordinary people and businesses in China, South Korea, and Japan that still use Windows XP. And ReactOS will be a great thing for those individuals. I can't really complain.

I only watched about 1/3 of it, but not a fan of what I'm seeing so far... as in, "Hey, let's remove all the controls and idiot-proof the interface." Take a look at the RSS example around the 5 minute mark; in the 'chrome' example they show, there's far more information on the screen at once, which I very much prefer.

And it seems like MS can't make up their mind. On one hand, they're like, "Let's create this new Metro interface and remove all the buttons and controls", and then they go and add the ribbon to Windows Explorer. Isn't the ribbon sort of a direct contradiction to the simplistic and minimalistic design of Metro?

Edited 2011-09-14 05:30 UTC

no.

Applications that are more complex in function than an RSS reader require a more complex interface...the Ribbon provides access to the more complex features of an application with out cluttering up the UI.

if history repeats itself windows 8 will be like vista where a whole bunch of changes were made and it will claim the title of vista2 or something like that. windows 9 will be like windows 7, a damage control release just before they break everything again with windows 10.

I don't think there's such a clear pattern in Microsoft Windows releases. Win-DOS was pretty much bad, bad, bad, decent (3.1); Win95 series good, good, shit (ME), NT series pretty good except XP and Vista really needing a service pack or two.

After Windows 7, they really need to start breaking things again, or they'll be stuck with it for another 10 years.

Yeah. Because Windows 95 was so good and Windows XP so bad?

Playing the devil's advocate...

Compared to its predecessors in the Windows 3.x family, Windows 95 was a huge leap forward. Compared to Windows 2000 which is its technical predecessor, XP was much more of a "meh" release.

Windows 2000 wasn't a consumer release. I never even saw a Windows 2000 computer in my life. XP was the first NT consumer release, and that makes it major; it's not a "meh".

Edited 2011-09-14 21:24 UTC

I bought a Sony Vaio ultra compact (for the time) with Windows 2000 pre-installed.

Yes it was.

2000 was not marketed as a consumer release, sold on everyone's computer, etc. But having used it at work like I would have used an XP box, one year ago... XP really was a mixed bag as compared to it, not big of an improvement and a bit of a regression.

Except for USB. USB support was a huge improvement in XP over 2000.

Played with the Windows 8 dev preview.

If you expect to be able to close applications, keep applications open, or arrange them on your screen, Windows 8 isn't for you.

If you're one of those people who never felt like they had any control over what their computer did in the first place, I'm sure you'll love it.


Couldn't get snap to work, possibly because no widescreen resolutions were available on my VM. Using the mouse was fatiguing (so much dragging). When you eventually figure out what's a button and what's just text, clicking said button can have any number of effects, none of which are predictable: easy example being the little Win+C/hot corner menu in the bottom left corner of the screen, it's items, once clicked, opening up a sidebar on the far right side of the screen.

one good thing i can say about it is that you can install bblean and have it as the default shell just as on windows 7 and all that metro stuff is gone for good (doesn't even autostart or anything, metro apps don't start up either probably because the runtime is not loaded and doesn't load automatically)

Good to know. Hopefully Metro will piss enough people off as to reinvigorate the alt shell community.

yeah, I was tempted to try it on VMs, but I don't have much time right now. I did notice that it requires a direct x 9 compatible video card. I don't think any of my vms on my linux box can provide that.

you assume a unified paradigm is the right way to do it. while I appreciate how small devices have emphasized the need to simplify application interfaces in general, I frankly don't see a way to create a unified interface for completely different devices with completely different uses. right now it sounds like windows 8 is ui wise a cluster. Apple and Google made the obvious choice to treat different devices as different and work on synchronization.

<sarcasm>
itunes will BE the OS.
</sarcasm>

Yes, this is exactly the same UI as the smashing success that is Windows Phone 7 which is crushing those pitiful, archaic icon based offerings from Google and Apple.

Clearly, everyone should just pack up and go home, Microsoft conquers all!

As in Vista and 7, press the Start button, then start typing to search for your desired app.

To pin an app, either swipe up or down on the app's tile, or right-click on it. A check mark will appear on the top-right corner of the tile, and an app bar will appear at the bottom of the screen. Click the Pin button. When you pin the app, its tile will be added to the end of the list.

Not in the Dev Pre build, but in the future, you'll also be able to create named tile groups to organize large numbers of tiles if you choose to keep many on your Start screen. See Julie Larsen-Green's demo in the Day 1 BUILD Keynote.

Thanks for the info!

they actually did some work on win32 as well with the new task manager/explorer (that's the obvious work obviously), I wonder what else they did.
If win32 apps continue to function as they do in windows 7 I might switch as I use bblean as my window manager anyway and it does work in the dev preview (there ARE start menu shortcuts, there's just no start menu opening from the start button, so bblean loads those just fine)

That's correct.



Nobody is forcing you to get rid of your code; especially if you already factored your code -- like many people -- so that there's a separation between how you render/composite your UI and how you get the bits to the screen. Most games will port with practically no issues at all. You can still use common languages (C/C++, C#) and new ones (HTML/JavaScript), as well. UI has to evolve.



THe windowing paradigm is still present. You can still access the desktop, and it works pretty much exactly the same except for the Start Menu. You don't have to use Metro or run any Metro apps if you don't want to.

They just bolted that other stuff on. Sure you can build Metro UI's in C++ but why the fuck would you want to. Metro is basically made for Javascript/HTML/CSS development. Their just throwing a bone to their base.

Yup. It sort of makes you wonder if they scrambled to do something dot net related after all of the complaints.

I didn't say it was dead. Did you read at all? I said it was on its way out.



No, it's not.



Read that link again, and focus on what it actually says, okay?



Someone clearly hasn't been paying attention. It's the other way around.

I didn't say it was dead. Did you read at all? I said it was on its way out.

Everyone on earth that knows anything about computers knows that Win32 and 32 bit in general is on the way out. We've knows that for about 10 years.

Metro is blatently a shell running on the same old gui.

No, it's not.

Mostly it is.

Metro apps can be made using the Win32 apip

Read that link again, and focus on what it actually says, okay?

Not only that, every single demo that we've seen so far has clearly shown a typical Windows setup with the Metro UI running as an application full screen, and when you run a normal native app it pops up in the old shell, right behind metro.

Someone clearly hasn't been paying attention. It's the other way around. "

MS has been busy lying again. Read my other post about what they promise and what they deliver.

Interesting link. It doesn't mean that metros can be created *completely* with win32 api. It means that metros can be created with *some* win32 apis in addition to the metro api.

Why do they allow that?

From the article:

Yeah, they didn't finish the metro api. It can't completely replace win32 in all scenarios. That's less than a clean break. They'll have to keep support for those win 32 api calls now for a lot longer than they would have otherwise.

Edited 2011-09-15 14:02 UTC

Heh. Yeah.

It's my opinion that Microsoft never wants to completely replace the Win32 API. If they did, if they made it possible to really write pure CLR applications it would make projects like Mono able to replace Windows.

No, they want to keep people dependent on P/Invoke for as long as possible, although with the new ARM support they are shooting themselves in the foot by requiring developers to recompile.

Developers can write pure .NET applications. They just can't use a lot of useful Windows features.

My particular feature was memory mapped files. It took them four versions of .NET before they got that in there.

I really wish I could understand why you keep saying these things. Watch Brad Linder's demo video on the Eee PC. Try the developer build yourself. OPEN YOUR EYES. This is BLATANTLY the same old win32 gui with Metro running as a 3d OVERLAY. The Metro style start button hovers over the area where the old start button was, and when you launch metro apps it just brings up the rest of the overlay.

http://webtrickz.com/how-to-get-back-old-start-menu-in-windows-8/

this is very good, and what you'd expect to see if they still plan to sell this to businesses. it remains to be seen how much of this bridge they'll burn before release.

Edited 2011-09-14 18:55 UTC

Honestly I am still shell-shocked that they think it is a good idea. What user interface textbook says you should regularly shuttle users between two COMPLETELY DIFFERENT IN EVERY WAY user interfaces. *zips back to Metro* Find that book for me. *zips back to Desktop* I dare you. *zips back to Metro* As a user I can tell you this is jarring *zips back to Desktop* like Where's Waldo projected onto my eyeballs with a strobe light. *zips back to Metro*

Edited 2011-09-14 18:52 UTC

I think it's ok to have two environents if you must, - but not the way this is done.

There are several ways this is commonly done today:

- The compatibility environment is a walled off bubble within the host. Settings within the guest environment do not propagate to the host, and integration is minimal (cut/paste and file sharing). The guess manages it's own windows and creates it's own desktop.

- Apps within the compatibility environment are re-parented (at least visually) within the host's shell. For example VMWare's "Unity" functionality. Still integration is only visual, and settings and configuration done using guest apps don't affect the host.

- The legacy binary is run via an API wrapper that maps API calls to the native equivalents and provides a graphic canvas to the app within the host environment, managed by the host's window manager. Wine is an example of this. A similar approach, requiring a recompile, is a compile time wrapper that maps legacy APIs to the new APIs (Apple's Carbon).


So this suggests some possible approaches for Windows 8:

- Semi-virtual environment for legacy apps with their own OS images and file system. Desktop is a separate app, and integrated only via copy/paste and file shares. Metro is entirely legacy free and totally un-aware of Win32 or the desktop.

- Same as the previous, but do away with the desktop entirely. Create a metro-ized wrapper for legacy apps that runs the apps within the Metro shell.

- Unified OS installation, no virtual environment, legacy applications are run via a modified win32 API that makes the apps "look" at least more native and makes them function properly within the Metro shell.

My preference would be the second. It keeps the legacy cruft walled away, but at least allows the apps to function within the native Metro shell as peers.

Haiku OS is what you want, every other desktop is leaving Tracker and basic OS minimalism further and further behind (sad in my opinion). Haiku with a few more improvements in Tracker like variable sized picture/content view like say ThumbsPlus and a few dozen apps would be enough to rock my world. We'll see.

No, you're definitely not the only one...

So I got myself the VMware Workstation 8.0 upgrade and installed the Developer Preview.

After tinkering around with it for several hours - together with some friends and colleges - and resisting the urge to throw my PC against the wall...

From my/our point of view Metro perfectly aligns itself with Gnome 3, Canonical's Unity 2D/3D and KDE 4 in its current shape and form. It's sooo utterly useless for getting work done(TM) or managing your PC that I'm/we're out of words. (@possible critics: Yes, I/we keep in mind that this is a unfinished Developer Preview which can/will change on the way to RTM - that's actually what I'm/we're hoping for)

A "classic desktop" containing the start menu as we know it from Vista/Windows 7 simply HAS to be added, otherwise power users and admins won't touch it even with a 10 foot stick.

Just my two cents on what we see at this moment.

So, all the desktop makers are committing suicide at the same time. nice. Didn't they ever hear of "If it ain't broke, don't fix it"

You also write...

Maybe it is broke? Now, perhaps not to you ...but then a) from the description of your desktop, you don't seem as a very typical user b) we did hear not entirely dissimilar wailing also when abandoning, say, paradigms of DOS era ;p (heck, or when going from 9x-style to Luna)

But, from numerous problems many people face with "computers" it seems they're nowhere near optimal for most of them. As for us?... I think we'll get through fine one way or another, we'll see (maybe partly a reintroduction of workstation concept? Hopefully not the price)

Edited 2011-09-21 00:03 UTC