If ever Microsoft did look in, they would have been asleep at 2.....I was.
"Type Hint" <> wrote in message
news:imsom2$ard$...
> I've seen a lot of weird Windows crashes in my day, but this one takes
> the cake. Definitely different from your run-of-the-mill STOP error.
>
> 1. I'm clicking around in an Explorer window and go to sort some folder
> by name. I think maybe Windows interpreted the click as hitting that
> "slide show" button above it, based on what eventually happened, but
> at the time all I knew was that the window didn't resort itself as
> I'd wanted it to.
>
> 2. I go to sort it again and it works. I do a couple more things. Then
> suddenly busy cursor and a moment later BAM! Blank black screen.
> Alt-tab does nothing. Ctrl-tab does nothing. Esc does nothing. Ctrl-
> esc does nothing. Ctrl-shift-esc does nothing. Nothing else does
> anything either. I have a mouse pointer and it moves but there's
> nothing to click on. Keyboard LEDs toggle but sluggishly and less than
> 100% reliably.
>
> 3. I'm sighing and reaching for the power switch when some other random
> Explorer window appears abruptly, relieving the blankness. I click in
> it figuring I might be able to use the menus to do *something* useful
> but the damn thing is not responding.
>
> 4. A few seconds later, there's a lot of automatic-seeming flickering and
> then my desktop comes back to normal -- almost.
>
> 5. For a little while, nothing responds, and then suddenly the start menu
> flies open all by itself. After that, the system's back to normal --
> almost. I cancel the unwanted menu and go about my business but soon
> notice that several applications are being sluggish. Typing in
> anything stops and then rushes to catch up, mouse clicks are ignored
> only for half of them to randomly take effect several full seconds
> later, and as if I'd just clicked even if something different is under
> the pointer now, etc.
>
> 6. I pop up ProcessExplorer and see that Exploder is chewing up 30-40%
> CPU.
>
> 7. I try to work on other things for a bit in the hopes that it will die
> down after a little while longer, as the system recovers from ...
> well, whatever the hell had happened to it to cause that blank black
> screen earlier.
>
> 8. Eventually, I get fed up and switch to ProcessExplorer. Right click
> ... wait 3 seconds ... Properties. Wait 6 seconds. Dialog appears.
> Click Threads. I want to know which specific Explorer thread is
> responsible for the excessive CPU use; perhaps I can safely kill just
> the one thread and not lose all my state in open Explorer window, and
> perhaps the thread's name will give me a clue as to why Explorer is
> spazzing, and perhaps that clue will let me despazz it, perhaps by
> closing some window I wasn't using anyway or something.
>
> 9. Instead, BAM! Blank black screen again. From clicking the Threads tab
> in the Properties dialog in ProcessExplorer this time.
>
> 10. Wait, what's this? One of my extensive collection of Kathy Ireland
> wallpapers. WTF? Who ordered *that*?
>
> 11. Then another, then another. Some sort of slide show? I have a mobile
> mouse pointer but no UI. Keyboard LEDs are stuck and ESC, clicking,
> double clicking etc. don't work. I even get desperate enough to hit
> F1. The one key I avoid is alt-F4, in case it goes straight through
> the slideshow that has hijacked the screen and hits some application
> I don't want to close.
>
> Meanwhile my mouse pointer is a hand in a couple of random spots and
> an I-beam in others, but an arrow most places, as if responding to
> the invisible contours of an unseen window *beneath* the one that
> nominally has the focus (the fullscreen app that's quite evidently
> hijacked my display).
>
> 12. It's unresponsive, other than that the slideshow doesn't freeze, for
> over five whole MINUTES. And then, suddenly, roughly as I was sighing
> and reaching for the power switch, a control overlay with pause,
> stop, etc. appears that should obviously have been there five minutes
> ago (if you ignore the fact that the slideshow itself should NOT have
> been).
>
> 13. Obviously, I hit stop. It takes a minute or so to take effect (natch)
> but I've got my desktop back! Only ... what the **** is THIS ****?
> Windows are cut off by screen edges and not where I left them. Other
> stuff is wrong. The whole goddamn thing looks blurry and low res. And
> there's a help window open, which was presumably the outcome of my
> earlier F1 keypress, but wasn't very helpful since it stayed under the
> "window" (if you can call it that) I needed help with and only became
> visible after I no longer needed any ****ing help from it.
>
> 14. I pop up control panel, noting that the system seems more responsive
> than earlier. I make a mental note to see if Explorer's CPU use is
> down out of the stratosphere in a bit, while I delve down into the
> display settings.
>
> 15. What the **** is this? 1024x768? How 1990s. Who the hell has been
> monkeying with my settings WHILE I HAD NO WORKING UI?!
>
> 16. Obviously, I change it back, then alt-tab to ProcessExplorer. Er,
> that is I TRY to alt-tab to ProcessExplorer. It's *gone*. Apparently
> that click on the Threads tab of the Properties dialog for the
> CPU-hogging Explorer process crashed the ****ing thing.
>
> 17. I restart it. Explorer's CPU use is normal (i.e., nearly nil).
>
> 18. WTFWTFWTFWTFWTF...
>
>
> Here are a few useful pointers for Microsoft that might come in handy
> when the time comes to start developing Windows 8:
>
> 1. One stray mouse click shouldn't render a machine nearly unusable for
> FIFTEEN ****ING MINUTES, unless it somehow does a Start -> Shut Down
> -> Yes, I Really Do Want To Shut Down, You Goddamn Machine, or
> something else drastic from that menu such as Reboot.
>
> 2. Clicks in Explorer windows, particularly, should not initiate anything
> that effectively grabs the whole system and locks you out of the UI
> for ages without there being some way to CANCEL the bloody thing
> quickly. In particular, Esc should cancel any fullscreen thingie
> that's not a game or something (and even then Alt-Enter should de-
> fullscreen the damned thing), and do it *promptly*.
>
> 3. Alt-Tab should always, always work, and alt-tab away from a fullscreen
> app should always, always minimize it.
>
> 4. Ctrl-Esc should always pop the taskbar and start menu up, even over
> fullscreen apps. Then you can always right-click the recalcitrant
> fullscreen app's taskbar button and "Close" it, or at least use
> Start -> Shut Down to get a graceful shutdown that saves all your
> stuff instead of maybe having to resort to power-cycling your goddamn
> box.
>
> 5. Ctrl-Shift-Esc should always start Task Manager, and if it's already
> started, force it to the front and focused, and if there's a
> fullscreen app that wants to stay in front of it, force always on top
> back on in Task Manager.
>
> 6. Likewise, F1-induced help should not pop under a fullscreen app, but
> appear over it, or the fullscreen app should de-fullscreen when you
> hit F1 in it.
>
> 7. Anything fullscreen that should exhibit screensaver-like behavior
> should disappear *instantly* if any key is pressed or the merest
> *twitch* of the mouse occurs (unless it *instantly* pops up the
> password prompt instead).
>
> 8. Exploder should not, under any circumstances, chew up significant CPU.
> It is inherently an I/O-bound task.
>
> 9. Nothing should ever trigger something that, when eventually canceled
> out of (somehow), will spontaneously *un*cancel itself five whole
> minutes later.
>
> 10. The slideshow feature's controls should not wait ten minutes to become
> available after the slideshow itself starts.
>
> 11. Things like ProcessExplorer shouldn't just disappear without so much
> as a "This program has performed an illegal operation..." messagebox,
> let alone an actual by-your-leave from the guy nominally in charge of
> the ****ing box.
>
> 12. The desktop resolution should not spontaneously randomize itself. Nor
> should fullscreen things like the slideshow hijack it but fail to put
> it back the way it was on exit. Nor should they default to something
> other than the desktop resolution the user set, at least not without
> a damned good reason, again unless it's a game and the user has
> explicit settings for that game that are dissimilar.
>
> 13. Mouse focus should always, always be on whatever opaque or translucent
> (not 100% transparent) thing is frontmost beneath the pointer
> position, and never be clicking or "feeling" *through* that to
> something further down the window stack.
>
> Put another way: Go down the video card's Z-buffer at the pixel
> position of the hotpoint. First non-100%-transparent pixel down that
> Z- buffer, find out what app it belongs to. That pixel of that app is
> what should receive the mouse events until that Z-buffer changes or
> the pointer moves.
>
> Taking notes? Good. Maybe you'll manage not to **** up Windows 8 too
> badly, then.
Similar Threads
Linear Mode
