Memo to Windows Vista

Oh dear. They haven't learned from XP, have they?

It got to the point where I'd have to remember to unplug my computers from the web before setting off any long simulations - it had never struck me that anyone would ship an OS that would randomly turn itself off in the middle of the night, when it's clearly doing something (you know, like 100% of the processing power being used).

I, of course, have never seen this happen (and I've been running XP and Vista with the download-but-don't-install setting forever). I bet you think you changed it, but you really didn't; it's been my experience that when people report unlikely problems, they're usually mistaken about what they've done.

Don't worry, Microsoft is working very hard to address the biggest problems with Vista: they hired Jerry Seinfeld to make funny commercials!

I, of course, have never seen this happen (and I've been running XP and Vista with the download-but-don't-install setting forever). I bet you think you changed it, but you really didn't;

You never encountering this problem carries exactly as much weight with me as the fact that I've never had a problem with an Apple product does with you.

I know I changed it, because it worked fine for the last couple of months. In fact, I remember it pestering me about the fact that the setting had been changed from the default, until I finally got that to stop.

My guess is that the last time I let it install updates, it reverted everything to the default settings again. That seems to be the pattern.

Please allow me to be the first to say:

"{ahem} Ubuntu!"

-Thank you.

I would also recommend Ubuntu. That is unless you intend on having any sort of working software with your computer, in which case you should avoid it like the plague.

In XP, you can stop the Windows Update service to prevent this sort of foolishness. Maybe that also works under Vista? Of course, Ubuntu woukd be a good solution as well...

I would also recommend Ubuntu. That is unless you intend on having any sort of working software with your computer, in which case you should avoid it like the plague.

That's true, all I ever do is stare at the splash screen.

/snark

I meet Stephen's /snark, and raise him /snark /snark
Ubuntu's the best, the only problem with updates is you have to enter the friggin password every time.

I vote for Ubuntu, as well, with the caveat that it does require more upkeep to make sure everything is working. It's like an old British roadster - you have to spend a lot of weekends tinkering in the garage, but when it's working, it's an absolute joy.

I recently had a conversation with Ubuntu about updates. I had clicked the icon and entered my password, and Ubuntu did the update. Then it popped up a message.

"Hey, to finish this update, I need to reboot. Can I do that now?" Ubuntu said.

"I'm busy now." I replied. "Go away."

"Okay, that's cool. I'm just going to stick this reminder icon in the taskbar. Restart when it works for you, and I'll finish your update."

Then Ubuntu shut up. It never bothered me again. So simple, and yet so absent from Windows.

Well, if you want to go to Apple, that's fine, of course. It's a wonderful operating system.

The quip about working software in Ubuntu seems a bit crass. Sure, a lot of software that runs on Ubuntu lacks some feature that you might find in Microsoft systems, but saying that there is no working software for Ubuntu is disingenuous at best.

Koffice, OpenOffice, and others produce documents just fine. So some of the nicer features of MS Office are not present (MS Excel features in particular), big deal. Work around it.

I'm the one who decides when you install and restart.

If you use proprietary software, there can be no complete third-party verification of what the software does, and you cannot decide. The algorithm software uses to interpret your commands is secret - you have no way of knowing when it will ignore you, or how it will respond. And if you're connected to the internet, you do not even know who else it might be taking instructions from.

I bet you think you changed it, but you really didn't; it's been my experience that when people report unlikely problems, they're usually mistaken about what they've done.

Please read:
Microsoft has begun patching files on Windows XP and Vista without users' knowledge, even when the users have turned off auto-updates.

This is a well-documented and empirically testable problem. If what you're doing must operate without interruptions, you'll have to do as Bob suggested, and either disconnect from the internet, or obtain the services of someone who knows how to configure a filtering firewall to filter out windows updates. Or you could try switching to another OS.

Ubuntu's the best, the only problem with updates is you have to enter the friggin password every time.

When I was young and naive I considered this sort of thing annoying. Now I'm older, and frequently used passwords are the only ones I can remember. So I've come to value regular requirements to use passwords.
P.S. I'm not regular ubuntu user, but I'm curious about the reports that it is hard to maintain, or requires a lot of attention. During the 6 months that I used it daily ( for C++ software development), I practically ignored normal computer maintenance issues, and encountered no problems.

Ubuntu does NOT require you to enter your password every time. You can set up autologin so you it will automatically login to your account after reboot.

One thing to remember with Microsoft, is that they are very willing to have their updates change settings. Whenever you do an update, you have to check the important settings.

Hmm, I've never had this happen on my Vista machine, and I often keep it running overnight. I have it set to download updates automatically and then inform me; I decide when to actually install and restart. I wish I could remember exactly how I had that set up.

Now, of course, I have other problems with Vista...

I've had an almost identical problem with Vista. As far as I can tell then problem is exactly as you say. Whenever I let it make a large update it resets its updating rules.

Two words:

Li ... nux...

Now together, as one word.

Linux.

Now two words again ...

Li ... nux.....

OK, now one word again....

Avoiding any more literary references to Ubuntu, it is possible to set the reboot time for Windows, after update, to something approximating the age of the universe so that it never interrupts before you turn your machine off some time for other purpose.

I'm using XP, and I've never had this problem -- though I think that my setting of "I decide what to download and what to install" was automatically changed to "everything is automatically downloaded, and I only get to choose what to install" every couple of months. The only annoying thing is that once it has installed all updates, a window pops up every 10 minutes or something and asks me if I want to reboot now, but, judging from comment 20, even this can be changed.

If you are running long simulations on a windows box you are "doing it wrong".
see: http://xkcd.com/463/

I used to run a huge number of simulations, and am currently doing some pretty heavy signal processing stuff. Matlab runs great on linux (ubuntu in my case).

If you are using someone else's simulation tools and they wrote them in such a way that they require windows, you are probably foolish to trust their code does what you think it does.

Little story:

A friend of mine was doing some pretty intensive spatial stats runs (population genetics work). 'Spaida' was the method, and he got a C program implementation from the original author of the method.

My friend comes to me one day, asking if I could look at the code and possibly parallelize it for him since it takes days to process his data on a single computer. I say 'sure', the algorithm described in the paper seemed simple enough and it would be fun to do an easy little programming project.

So I sit down with the author's C code and his paper. The code is a horrible mess, but I tease it apart into a more sane structure (verifying it produces identical output along the way.) Not only was the program messy, it had stupid inefficiencies, and worse still, didn't correctly implement the algorithm! The program's author was the same guy who wrote the method/math... and he got it wrong in the program.

So I cleaned up the code, fixed the math, and got rid of some of the most glaringly stupid inefficiencies. >100x speedup without bothering to parallelize it... oh, and it actually implemented the method properly. I did parallelize it, but don't thing my friend ever bothered running it on the beowulf cluster since the serial version was 'cup of coffee' fast running on his desktop.

I have several more nearly identical stories involving different methods. Probably the worst was an 'optimal foraging theory' program, which I just reimplemented from scratch (original was in BASIC!) and converted from O(N^3) to O(NlogN). Of course, adding in little things like a parameters file (instead of modifying the code for each different simulation) and standardized output suitable for automated analysis (instead of having to cut and past from a messy verbose stream of messages into a spreadsheet).

Anyway, my short form... If you are doing simulations or computational models, and don't know exactly what the code does, you should probably write it yourself. There are a few exceptions, but those are either standard enough that multi-platform implementations exist, or else written by people you are directly collaborating with.

Regarding difficulty maintaining Linux... it depends on the distro.

For Ubuntu, the 'LTS' (long term support) versions are pretty easy. The non-LTS versions are more geared towards development. I'm still running 6.0something on my desktop at home without any trouble... though I'll upgrade soonish.

RedHat has a similar scheme I can't remember off the top of my head. Plus they have 'Enterprise' versions which have a slower (more conservative) update cycle and longer term support.

CentOS is specifically designed to be long-term stable with respect to updates. We are using this on a DB server, but are going to switch over to the latest Ubuntu LTS as we upgrade the actual hardware (just familiarity and aesthetics really, I like Ubuntu better).

The main problems people run into with linux and installation (which has gotten pretty easy for most cases) and getting some software up and running properly. The 'problematic' software tends to be big stuff dependent on lots of little stuff like databases, web frameworks, ect... which really take some effort to setup properly regardless of the OS. Most big distros now have 'virtual packages' to install and setup a reasonable default configuration for the most common 'big stuff'.

In linux you may get messed up with software configuration options... but windows will either have the same potential stumbling blocks or not have the options at all.

Jon @ #10,

Yes, but could you get Ubuntu to make a decent cup of tea?

This all reminds me of a few weeks ago when my XP at work asked me if I wanted to do updates. Knowing the unfortunate course of events which would follow (foreshadowing here :) ), I clearly and emphatically clicked "NO". Rebooting the next day, I discovered that XP had cheerfully ignored my request and now I had the updates, effectively breaking company software on my laptop (don't get me started on how that shouldn't have been a problem in the first place *grumble grumble*).

I'm actually using SuSE 10.3 at home, which has the unfortunate habit of needing a kernel parameter at boot to not break the network card driver *sigh*, but that's another story. Upgrade later..though they seem to have been going downhill since Novell bought them.

Ever since that animated paperclip thing, I've advocated the installation of a taser into bill gates' rectum.
I've never had this particular issue with updates; my machines are behind a firewall with managed updates from our servers and I wonder if this only affects home or small office users.
Would you rather your pc was zombified?

RE #22:

Great story. I wonder how many of the original author's conclusions about the model were wrong because it was never implemented properly.

Amusing story:
There was this mildly famous program for simulating collisions, known as the X Method. Guy who is paying attention finds a major error in the code that invalidates all of the work done with it. The corrected version of the code, which actually implements the X Method as it was originally described in the Physical Review, is then called Method Y. The paper that describes Method Y used just enough obfuscation that the referees did not realize that it was actually the X Method, and that the cool new results of Method Y were simply from a correct calculation.

There are probably more instances of this than one cares to know about.

Ahem. I am a fairly dedicated Ubuntu user, but seriously, you people are telling Chad some bullshit. It does not Just Work.

The most recent episode was when I had some side work helping a CS student suffering from massive carpal tunnel work on his diss. The work was being done on a Dell laptop, no compatibility issues, no problems at all. For two weeks the results for various timing tests were solid since we didn't touch the features tested. Then one day I come in and he tells me that things have gone haywire. The tests are running nearly twice as long on average but no stable time like we used to have. I ask if he's done anything, he can't recall anything, so I check menu.lst. Sure enough, grub has kernel entries that weren't there the day before -- somehow Dapper ignored the do-not-update setting, on a machine being used several hours a day for development and testing. These timing results must be absolutely reliable and comparable from day to day, so we had to scramble to find another machine for testing, eventually getting access to some Macs (which were unsurprisingly rock solid since their software profiles are not changing unexpectedly).

So seriously, my anecdote as a dedicated Linux user is worth the same as anyone here. Now, if only f'in X wasn't such a pain.

My old rule was never buy a Microsoft product that's less than a year old. How old is Vista now? I think it was released over a year and a half ago, so I've updated that rule to two (with further updates to come).

My old Linux machine recently fried its power supply or motherboard. My son (until he goes back to law school next week) is sleeping in his room with the door shut, barring me from his Windows NT maxchine souped up for fast game play. So here I am on an ancient Windows 98. At random times, averaging once per half hour, I have to shut down some bogus task (sometimes a task called ) which is sometimes the task Aupdate. Sometimes it tells me that I'm overdue to update virus filtering. Sometimes it just freezes, leaving me unable to save my partial results. Ghod, I miss using that Linux machine. Oh, and my ancient Mac has memory too small to update the browser, or email, and so can't do either. Fortunately, my slide rule never updates itself. But it doesn't word process worth a damn.

And while we're ranting about automatic software updates....

Dear Adobe:

When I start one of your programs, it is because I want to use it, and I want to use it now. I don't want to wait umpteen minutes while you go and update some other component of the software, and not let me run the program I want to run while you are busy updating.

Apple has a much better approach: you can set up a schedule for when it checks for updates (or manually tell it to check, an option I use on the home machine where my internet connection is not always active), and it does not actually download and install the updates until you tell it to do so.

Would China have ever conquered the world if the abacus kept stopping in mid-computation to change bead versions? Would Gerbert ever have become Pope Sylvester if his introduction of the abacus to Europe had been incompatible in character set?

Zombified. Good word. Just saw "I Am Legend" again, with my wife, on DVD. One does not, of course, innoculate to CURE a viral epidemic, but to prevent infections. Still, Bill Gates comes across like Dr. K who opens the 3rd film of that classic fiction with an interview on TV where she modestly admits to curing cancer with a genetically engineered virus, and proceeds to make a dreadful dumbed-down metaphor about replacing bad men in cars with cops. So, in conclusion, Bill Gates and bad science reporting both lead to the destruction of human civilization.

"I vote for Ubuntu, as well, with the caveat that it does require more upkeep to make sure everything is working. It's like an old British roadster - you have to spend a lot of weekends tinkering in the garage, but when it's working, it's an absolute joy."

I was walking around town some time ago and saw a bumper sticker on an MG, "All parts falling off this vehicle are of the finest English workmanship". Never let it be said that the Brits don't have a sense of humor.

On topic: Forced updates! Yikes! Get a Mac. (Then you get your choice of THREE idiosyncratic operating systems).

I'm pretty sure the truth of the matter is that XP is the pinnacle of Microsoft, and is at worst one of the best three OSs of all time. I know, it's easy to give MS products shit. But the thing simply works. It's rock solid. I haven't restarted my computer for months. I'm pretty sure I have nowhere else to go from here, unless I find myself with unbelievable amounts of free time and decide I want to piss it away learning Linux (/bin/usr/recompile kernel) or have an extra thousand bucks jingling around in my pocket and want to buy a Mac. So it's me and XP, for now and the foreseeable future.

Recently got a new computer, which came with Vista pre-installed. I stuck with Vista for a week; at the end of that week, I made backup disks of everything, formatted the hard drive, and installed Mandriva. Bliss!

Vista has been out well over a year now, but it is the most bloated, inefficient, inconvenient, user-unfriendly pile of crap I've had to deal with since, well, Windows ME. And in a few ways it's much worse. MS seriously screwed the pooch.

If you really want Windows, stick with XP; there is no discernable improvement over XP that I can tell, and quite a few things that Vista does so badly that it goes all the way through "wrong" and out the other side into "f***ing awful".

Do consider Linux as an alternative, however. I like Mandriva, not just because it is based around a stable Red Hat kernel (which I am familiar with), but also comes with a very easy installer and user-friendly interface. And, oh yes, you can run a Windows emulator on it which seems to allow all my Windows-only software to work. The software may not work as fast as it would on XP, but there is not a huge difference, and from my benchmarking is actually faster than on Vista, disgusting bloated memory-hog that Vista is. And for the software that has Linux versions (most of what I use), I can actually get a worthy performance out of the hardware for the spec it is. With Vista, I had a machine with two dual-core x64 2.00GHz CPUs and 4GB memory, and it was running like a basic pentium simply because it didn't have a high-end graphics card to go with it and it was sucking all the system's power just to run the pretty desktop.

Basically, Vista can just bite me. I will not use it.

God, please don't listen to the people recommending Linux.

Re #35

If there was a god worthy of the name, he/she/it'd probly be running this universe sim on a massively parallel supercomp with various posix compliant flavours for its various susbystems.

It'd still crash every 30 gigayears or so.