Discussions

Congratulations! The product may have shipped, but the work sure isn't over yet.

15 hours ago, DCMonkey wrote

*snip*

It doesn't seem to have caused much of a problem with Windows 7's taskbar thumbnails, which show a close button when you hover over the tile.

Try using those on a touch screen.

3 hours ago, DeathBy​VisualStudio wrote

*snip*

So you're going to give the Windows Store app IE guys a pass then huh? Nice.

I've already managed to accidentally close IE tabs instead of switching to them precisely because of this, it's a bit of UI I really don't like. I'd have gone for a more Semantic Zoom type approach, maybe arranging the pages like the old Quick Tabs functionality from IE8.

*snip*

Your working under the premise that an app centric world is the preferred. That premise is flawed. People want to get to their data as quick as they can and don't give a rat's arse if that's by opening the data from a menu from the shell or from within the application itself so long as it's the fastest. Of course you'll always prefer the apologist's position.

App-centric seems to be working rather well. It's been embraced by every smartphone, tablet etc ever since Apple proved it made life easier. Of course you think anyone who doesn't agree with you thinks there is something that needs apologising for.

*snip*

With what -- only 2000 apps in the store? Maybe? Are you sure you want to go there? Looks like Citrix has a release out. I bet that's going have some pain points just like remote desktop. Can't wait for the others...

I didn't say it had to be a store app. Could be anything.

Come to think of it I have an app that would suck as a Windows Store app because of this very issue. It's an app that displays real-time weights from multiple scale heads. It's a WPF app that uses WCF with WSDualHttpBinding for communications providing both the real-time weight values from the scales and a control interface for the scales.

Having an app that's not written in a way that's connectionless isn't the same thing as an app that can't be.

*snip*

It's sad to see you make up lame excuse after lame excuse for Microsoft. You really think a full blown Windows Pro tablet is going to be so much more optimized than the Build tablet that it will meet or exceed the iPad or an Android tablet on startup times? From sleep even? Dude, put down the crack pipe. It's not helping you.

I think Windows SoC devices (whether ARM or Intel) will be a lot more comparable to other SoC devices than traditional old-school PCs. I think the UEFI systems will startup faster than BIOS-based machines. And from sleep? Do you honestly think even current PC's have trouble resuming from sleep?

Do you honestly think the BUILD tablet was even nearly representative of the hardware that will be launched alongside Windows 8?

"W8 grade hardware". Sounds like you're calling Sinofsky a liar considering he said we'd have 450 million customer available to us as launch day -- you know those upgrading from "W7 grade hardware."

I think you should go listen to what Sinofsky actually said at BUILD. Or, you know, keep making stuff up that fits your argument better.

Hey I hear Apple is looking for a new PR guy for their mapping app. I think you'd be perfect for it. You can use me as a reference; you'll get nothing but high marks from me.

Nah, they need someone who lives in his own little world, ignores what most people want from computers and will endlessly shift arguments every time it becomes obvious what he says makes no sense.

I'll pop your reference in the post, then.

14 minutes ago, DeathBy​VisualStudio wrote

*snip*

If the concern is that the "x" may be easily (not confusedly) selected then they could hide the "x" until the mouse hovers over the tile. Oh but that would go against the great Sinofsky who once said "Write for touch, F the mouse & keyboard".  I think that's what he said anyway.

Yes, that is the concern. And the absolute most ridiculous way to try and prevent it is to have that button be hidden until you mouse over where it is, because then you're almost certainly going to have the situation where someone moves the pointer over what looks like the switch target and has already clicked the button before realising the magic close button has suddenly appeared beneath their cursor and they've just quit the application they wanted to use.

Of course you could alleviate that problem by having the user explicitly right click with the mouse to make the close option appear...

*snip*

Um no... It's providing a quick route to a file I had recently created.

In what sense is that not file centric? And it still ignores the way Jump Lists are implemented, which is fundamentally around providing launch parameters to apps.

*snip*

I'd think that any type of terminal app would fall into the same category as remote desktop. That would also go for any admin tool that monitors and reports health of web sites, server, etc. Oh here it comes...I can feel it: those kind of apps are holding it wrong.

So that's a definitive "No." to the question, can you name anything other than a Remote Desktop (or equivalent) application then?

And yes I'm ignoring admin tools that monitor sites, because most of the widely used ones in existence (such as the excellent Nagios), are already implemented as web pages and so clearly capable of operating in an entirely connectionless fashion.

*snip*

Also I'll just ignore the delay it takes for me to power up my W8 tablet from sleep and get to the app I want vs. doing the same on an Android or iPad too.

I'm impressed you have W8 grade hardware, given that none has been released yet.

@JohnAskew: There is either something in there very cool that they're trying hard to keep back till the last possible second, or they're taking full advantage of all the speculation to build up a certain amount of hype (which is actually kind of working judging by the amount of WP8 speculation around).

I suppose the last possibility is that it's an agreement with existing WP7 supporting carriers to try and prop up availability of new WP7 compatible apps up to and beyond the launch of WP8 (after all, they are still selling those handsets right now). I think that's a little too short term to be a viable justification though.

1 hour ago, evildictait​or wrote

*snip*

I'm not sure Turing says that at all.

In fact, I think it is entirely possible to have a virtual memory system that never page faults.

For example: The non-paged pool. Also Win32k's shared session data.

I assume he was making the common mistake of confusing "virtual memory" with "swap file", which is so ubiquitous these days that it's rarely worth correcting. It's obviously possible to have virtual memory without page faults, Windows CE for example does it across the board (it doesn't even support swapping).

What if you allowed for multiple calibration points and then, behind the scenes, applied typical image distorting algorithms to make it fit? There has to be a better use for those than just playing around with making funny faces, right? Couple it with some image analysis to try and determine "paths" that a user followed as they walked about and you could be well on the way to something seriously awesome.

Obviously for entirely stylised maps (Like the London Tube) it'd be useless, but even though tourist sign maps aren't always to-scale, they probably mostly accurate orientation-wise.

1 day ago, DeathBy​VisualStudio wrote

*snip*

I disagree Andy. I think it does makes sense to have close buttons and good mouse support and jumplists because that's what desktop PCs needs and that's what people say when I plop them in front of W8. W8's new metaphor for applications life cycle et al is for devices. No matter how you try and tap dance around it W8 is a very glitchy, unfinished OS. It's a grand beta test that only a true Microsoft apologist can pile on enough "imagination" to deflect from its detractions. You know stuff like this gem:

Under normal circumstances you don't need to close apps, so putting a close button on the switcher is just begging for someone to accidentally "click" the close button rather than on the main switching part. It what sense does it make sense to design a UI such that bringing the app you want to the foreground is easily confused with closing it?

As for Jump Lists, they're very much a file-centric/launch parameter based approach to trying to provide a quick route into an application. When you aren't launching an app, but instead switching too one, they really don't make any sense. Of course an app being able to provide such a quick route into specific functionality, for those who want it, is very handy - so much so that Windows 8 apps can perform exactly that via Secondary Tiles and in a way which is much more flexible than Jump Lists could ever be.

*snip*

Really? A few cases like remote desktop? Oh and yeah if they'd only improve those pesky protocols...

Yes, really. Aside from Remote Desktop (or an equivalent) how many actual scenarios are there in which you are so fundamentally reliant on a connection to a remote machine that they can't possibly restore a usable UI state before a connection is restored? Can you name even one?

As for improving how it works underneath, yes that can help too. I'd put good money on Citrix doing a much better job when they release the inevitable GoToMyPC Windows 8 app.

*snip*

Oh yes because it's not a problem on your uber PC with tons of ram it certainly won't be a problem on Windows RT tablets with much less capable processors and a lot less ram. And if it a problem well they just must be holding it wrong.

You're argument is all over the place (as usual). If Windows 8 and the Metro approach is, as you say, only good for mobile devices, then it won't be any more of a problem for Windows RT tablets than it is for iPad users. If applications being suspended is as issue for such devices, then naturally it'll also make the iPad (which does the same) a complete failure too (hint: it wasn't) And if it's desktops you're worried about, then the spec of my current PC (which is a laptop) isn't exactly anything stellar, so it certainly won't be over the lifespan of W8. Nor is it necessarily a sign that it'll fail to work on a less capable device either, Notepad on this machine works just as well as it does on a ten-year old PC.

@Sven Groot: It's less about how long it takes to save/restore state, so much as how quickly you can get back to feeling "usable" again. It's a matter of reducing state restoration on start to exactly what you need to get back to a workable state quickly, pulling any remainder in asynchronously or on-demand as required. There are very few cases where it's quite as pathological a problem as with Remote Desktop (and even that could probably do better with improvements to the underlying protocol).

I'm not much of a Pinball player but I can't actually get it to do that on my machine (despite launching every Office app, Zune, Visual Studio and having a few games of Minesweeper and Freecell in between). I've got a lot of ram in this machine though, so it's hard to say whether that's atypical. I just find it switches to the "Paused" screen, which seems sensible enough, though I think it'd be less jarring if it was a paused state showing the game in-play and could be un-paused by pressing one of the keys. I'd completely agree about the intro sequence though, it's one of those horrible advertising travesties that only games publishers seem to think is a good idea. Doing that is very much against the spirit of Metro and in the worse case scenario, leads to the kind of horrible experience you're seeing.

@DeathByVisualStudio: And when did I say it did? I explicitly said the sidebar thing is not a Taskbar but more akin to an MRU list. And that trying to bodge on things like Close buttons or Jumplists in order to turn it into one doesn't make sense. I know you love to disagree with everything I say just because I said it, but a least make some effort to be consistent, eh.

@Bass: It's not just about killing apps, it's about the entire process lifecycle. From what initiates it to how it has to behave whilst 'running' right through to how it ends (regardless of whether voluntarily, at the users request or at the behest of the system itself),

@Sven Groot: Which is why Remote Desktop makes little sense as a Metro app and, as far as I can tell, only exists because Windows RT devices might need to connect to other systems remotely. It's not because the principle itself is unsound, but more because Metro tackles the "low hanging fruit" part of the problem and deliberately avoids the more challenging areas (which is simultaneously why you can't run Metro apps windowed, because you reintroduce the problems that were being deliberately left to one side till a future version)

There's no reason that in a desktop environment where application lifecycle is managed by the system, you can't maintain some semblance of "state" based on the fact an application has a window open, indeed to some degree you need to anyway (much as Metro would never suspend the foreground process). That doesn't mean that it's necessary to start every application manually (so Messenger could still inform you of a new IM without you ever having to explicitly end it and Outlook would always pop up your reminders regardless of whether you remembered you need to open it to check) Nor does it mean that closing the last "open" remote desktop windows has to actually terminate the underlying process, the OS would be free to keep it "hot" if it didn't need to relinquish the resources in case you want to use it again. The point is to free the user from unnecessary resource management, not to prevent them saying "these are the Tasks that are important to me right now"

Page faults are fine. Swapping to disk is fine. Being forced into a situation in which a suspended application is terminated is also fine. Sometimes things will happen that the OS didn't predict as well as you'd have liked and it takes a tiny bit longer as a result. This is all as true today on Windows 7 running traditional desktop applications as it ever will be. As the slower bits of computers (IO mainly) get faster, it just becomes less and less of an issue.

I think you're massively over-complicating how it actually works; the OS doesn't ever have to reason about execution, it simply keeps track of what state a process is in and whether it therefore needs to schedule any CPU time for it. Likewise it responds to requests for memory or IO as and when they happen (ignoring for a moment that memory managers often pre-empt expected future requests). 

There is no difference in resource management for an OS in which all processes are conceptually always running than one where you've manually launched them all and just left them open. The OS already has to do the difficult balancing act of moving things in and out of memory, allocating CPU /GPU shares, balancing IO contention etc. It doesn't get any more difficult because you add other processes that are conceptually running but not actually demanding any resources at that particular time, because things not demanding any resources are a non issue. Even on Windows/Linux/MacOS today the OS itself doesn't actually care that a process is "launching" as opposed to just moving from the "waiting" to the "ready to run" queues.

It certainly matters to application developers, because it changes many of the assumptions about how it's acceptable for a process to behave and particularly what it means for a process to start or stop (there is a much greater emphasis on getting back to where you left off for example). And it matters to end-users because it relieves them of having to do things they might otherwise forget (ye olde "I didn't get the reminder because I forgot to start Outlook problem"). The OS itself (whether Windows, Linux or Mac) has been treating the world as if that's how it worked for a very, very long time now.