Adobe installed an update...
-
@joe.edwards said:
There are easily thousands of ways a program can cause mischief once it's on the other side of the airtight hatchway.
I think you missed his point about the DOS attack. Because it doesn't require crossing any hatchways, airtight or not.
Hell, remember back a few months? Eclipse did (what was effectively) a DOS attack on my user directory without needing to break any airtight hatchways. Allowing this is a bad thing which the OS should attempt to mitigate.
-
@Mason Wheeler said:
And in 1991-ish, you didn't run 37 copies of the same program. (Live by the context, die by the context.)
Ok? What's your point? Microsoft shouldn't proactively fix flaws regarding things people don't generally do? Did you spend even 3 milliseconds thinking about the response I'd make to your little sentence there?
@Mason Wheeler said:
Perhaps, but what I remember is that I stopped running programs off floppy discs back in the 286 era. Floppies were for installing stuff onto the hard drive, or for transporting files from one computer to another, but certainly not for running programs when you had a hard disc available! That's just crazy talk, even way back then.
Ok, well when Microsoft makes Windows: Mason Wheeler and ONLY HIM AND NOBODY ELSE Edition then your argument here will hold some weight.
To spell it out: nobody gives a shit what you personally did. Windows already had hundreds of millions of customers by the time this design was being done.
@Mason Wheeler said:
I wouldn't know. Like I said, I'm not a Linux user. That doesn't mean I can't call Microsoft out on bad design decisions when they make them.
So do I. I have several times in this very thread.
@Mason Wheeler said:
And what does the Registry have to do with "central management of networks" anyway? Because that's something else I'm not: a network administrator.
Then you're not qualified to offer an opinion. Again: WINDOWS IS NOT MADE SPECIFICALLY FOR YOU.
@Mason Wheeler said:
Maybe if you're a network administrator. Maybe. Again, I wouldn't know. I'm not a network administrator. And furthermore, the vast majority of users are not network administrators. And the vast majority of programs that get installed on Windows systems have nothing to do with network administration. So if you're going to try to justify something that screws things up for everything, you'll need a better explanation than "it makes network administration easier" if you want to be taken seriously.
Except it doesn't screw things up, so that doesn't fly.
"Making network administration easier" is kind of the entire purpose of Windows NT. So... it kind of is a pretty good explanation, at least when it comes to any NT-based OS, or any OS designed primarily to share a network with Windows NT.
-
Maybe it's my fault for having e-mail subscription to this thread, but when I get home to 45 alerts, it's a sign the entire thing is out of hand.
It may be rich coming from me, but everyone just shut the fuck up for once. Half of the posters don't even know what they're talking about, or should I say they know enough to be cocktastic egomaniacs, ignoring everything those with the genuine knowledge have said, or twisting it to fit some ideal that they've already decided upon.
I'm going to leave one last message, which is to be considered my reply to any and all following posts: GIF Sound the Video
-
-
@blakeyrat said:
Maybe the 0.00001% of users who do retarded things or have retarded network administrators got it. But "users" as a generalization were not affected in the slightest.
Well, it seems that some people were affected line Lorne because of a slightly different issue caused most of the time by OEMs like Dell or HP in some models of laptops where Skype was either preinstalled but not activated or partially uninstalled however it goes without saying that MS should do better QA on the updates from now on to prevent this. Thread about this with some really insane comments about how all of this was a cyberweapon test
-
@blakeyrat said:
@Mason Wheeler said:
And in 1991-ish, you didn't run 37 copies of the same program. (Live by the context, die by the context.)
Ok? What's your point?
My point, which I made a few posts ago, is that the Registry was designed as a central registry for COM servers, which was a genuine global issue that needed a global solution. It was never intended to do what they did to it in W95, which is why it's so badly broken now.
And my other point, with regards to multiple instances of programs running, is that when your hardware gets better, it tends to get better across the board. A computer can handle multiple instances today, when it couldn't earlier, and also has fast enough disc access to handle configuration changes more quickly today than in the past, when it might have been an issue except that it wasn't because you weren't running multiple instnaces of the same program due to more restrictive hardware.
Seriously, do you have any concept at all of "reading comprehension?" I've explained this about as plainly as I can, and you seem to just not get any of it. Do you have some sort of mental bug where anyone disagreeing with you throws an unhandled exception somewhere in your natural language parser?
@Mason Wheeler said:
Maybe if you're a network administrator. Maybe. Again, I wouldn't know. I'm not a network administrator. And furthermore, the vast majority of users are not network administrators. And the vast majority of programs that get installed on Windows systems have nothing to do with network administration. So if you're going to try to justify something that screws things up for everything, you'll need a better explanation than "it makes network administration easier" if you want to be taken seriously.
Except it doesn't screw things up, so that doesn't fly.
Right. I forgot. That whole darned unhandled exception thing again. You refuse to believe it despite all the evidence to the contrary.
"Making network administration easier" is kind of the entire purpose of Windows NT. So... it kind of is a pretty good explanation, at least when it comes to any NT-based OS, or any OS designed primarily to share a network with Windows NT.
It was the purpose of Windows NT, waaaaaaaaay back when it was called "Windows NT" and there was a completely separate consumer-grade Windows alternative. (And back then, it didn't have the Registry as anything beyond a COM registry either, at least not before 1995.) But today, when Windows NT has absorbed the consumer-grade Windows into itself and it's now just called "Windows", it's kind of stupid to look at the whole thing from a network admin-centric viewpoint.
And that still doesn't explain why they thought it was a good idea to do something that, according to you, exists primarily for the benefit of network admins, in the consumer-grade, non-NT Windows 95. Seriously, if you can't keep the most basic details straight, how do you justify berating anyone else about their knowledge of the history of the product?!?
-
@RaceProUK said:
Maybe it's my fault for having e-mail subscription to this thread, but when I get home to 45 alerts, it's a sign the entire thing is out of hand.
It may be rich coming from me, but everyone just shut the fuck up for once. Half of the posters don't even know what they're talking about, or should I say they know enough to be cocktastic egomaniacs, ignoring everything those with the genuine knowledge have said, or twisting it to fit some ideal that they've already decided upon.
I'm going to leave one last message, which is to be considered my reply to any and all following posts: GIF Sound the Video
You're right. We should all quit this thread and go over to this thread, where there is tons of geniune knowledge being revealed and shared!
-
@blakeyrat said:
@joe.edwards said:
There are easily thousands of ways a program can cause mischief once it's on the other side of the airtight hatchway.
I think you missed his point about the DOS attack. Because it doesn't require crossing any hatchways, airtight or not.
The whole idea of the metaphor is that the battle for security is already lost once the program is downloaded and executed. The wall does no good once the Trojan horse has been brought inside.
Though I admit when I see the word "attack" I was thinking something done deliberately to be debilitating, which is why I felt compelled to point out that there are so many other ways to achieve that goal. Buggy program behavior doesn't fall in the domain of security (though security can limit the possible damage), but an "attack" certainly would.
Is it the job of the operating system to mitigate bugs in programs? I understand once a buggy program is shipped it must be supported, but if the bug isn't gracefully covered up then maybe it would never have shipped in that state.
-
@joe.edwards said:
Is it the job of the operating system to mitigate bugs in programs? I understand once a buggy program is shipped it must be supported, but if the bug isn't gracefully covered up then maybe it would never have shipped in that state.
IMO if there is something the OS can do to encourage/force correct software, it should. An ideal system would make incorrect software impossible.
-
@Mason Wheeler said:
And that still doesn't explain why they thought it was a good idea to do something that, according to you, exists primarily for the benefit of network admins, in the consumer-grade, non-NT Windows 95. Seriously, if you can't keep the most basic details straight, how do you justify berating anyone else about their knowledge of the history of the product?!?
Windows NT allowed windows 3.x to have roaming profile support, well simplified it a bit. You could have each user have their own personalized version of windows before NT, but that was a bitch to get working.
-
@blakeyrat said:
Many but not all. This is something the OS should do, you shouldn't have to rely on developers to provide it. (Especially since developers are shit.)
Developers still need to determine what the user should and should not have access to. The difference is the implementation.@blakeyrat said:
@nonpartisan said:
Because you've not seen the Registry corrupt in 2000 or later, you don't think it needs fixing.INI files can suffer a DoS . . . this is bad if the file is holding security information: Technically possible, I've yet to see it, if you've got a program misbehaving that badly you've got other issues.
So you admit it's a problem but since you've "yet to see it" you don't think it needs fixing.
I concede nothing on this statement. Any incorrectly-written software, given the right permission level, can cause a problem with other programs.
@blakeyrat said:
@nonpartisan said:
Again, I concede nothing here. Separation of large amounts of binary data and other configuration information is a sane practice.INI files can only contain strings; binary data needs to be encoded: for short amounts of binary data, encoding is not a big deal. Powerful systems that they are these days, the difference in processing is miniscule. My known_hosts files for SSH encodes the key for a known system, so what? If there's that much binary data to be stored, it should be stored in a separate file anyway.
Well the Linux world already fellates text data and tries to pretend any other type of data simple doesn't exist. So I wouldn't expect you to understand. But in an ideal world, yes, the configuration data would be able to store images, sounds, whatever. I don't see how there could be any real debate over this.
BTW, this is another "I admit it's a problem but somehow I don't think it needs fixing".
@blakeyrat said:
Remember the context of the post. The move from INI to the Registry was done on 386 and 486-class machines with 4 MB of RAM.
Be that as it may, Linux distributions were still using text-based configuration files. They still use them today because it's a proven way of creating a system configuration that doesn't have a single point of failure.
@blakeyrat said:
@nonpartisan said:
We agree on this.Many programs open the INI file and read them directly so the format cannot be extended: . . . why??? System settings are under /proc these days. There's no reason a program I'm running should be looking directly into the configuration files, save for the program itself. And since Linux distributions do not use .INI files you can have them in any format you need.
Linux doesn't have this problem
@blakeyrat said:
So you're argument is fine as long as you don't want any Linux OS to ever been centrally-manageable. But a lot of companies do.
I EXPLAINED TO YOU HOW IT CAN BE DONE!!! IF YOU'RE NOT GOING TO READ MY FUCKING POST I'M NOT GOING TO REPEAT IT BLAH BLAH BLAH DE BLAH DE BLAH!!!!
@blakeyrat said:
And then how do you prevent users from changing configuration? Read-only permissions? Now what if you want to let the user change their homepage, but no security settings, in the browser? How do you do that in Linux? Is it even possible?
Wow. Group Policy is not a panacea. The applications need to be written to honor what Group Policy is sending out. If I write a browser that completely ignores the GP-configured proxy settings, guess what? The proxy settings aren't going to be used. Similar accomodations could be made under a Linux system very easily if a developer wants to do that. Have a custom configuration file in the user's home directory, read that, then read the system-configured /etc config file so that it overwrites any user-customized settings that aren't supposed to be changed. Problem solved. Administrator can update the /etc configuration file with mandatory settings. Whatever a user customized improperly will get restored by the administrator's config.As for arguing about the .INI format and ASheridan's response, you lit the fire by linking to how the .INI file format sucks before he said anything. When that link doesn't even apply to Linux. Because Linux never used the .INI file format. Epic fail Blakey.
-
@nonpartisan said:
Wow. Group Policy is not a panacea. The applications need to be written to honor what Group Policy is sending out.
All they need to do is properly use the Registry in the intended manner. Now it's true a lot of applications fuck it up, most of them open source applications from my experience, but it's not difficult.
@nonpartisan said:
As for arguing about the .INI format and ASheridan's response, you lit the fire by linking to how the .INI file format sucks before he said anything. When that link doesn't even apply to Linux. Because Linux never used the .INI file format.
Yeah; what it does is WORSE. Because at least INI is a standard.
-
@nonpartisan said:
Be that as it may, Linux distributions were still using text-based configuration files. They still use them today because it's a proven way of creating a system configuration that doesn't have a single point of failure.
Wow, I'd like to see what distro you use if it doesn't have a single hint of GNOME, and therefore, GConf being the one true way to store configuration. Must have been hard finding a popular browser, since both Firefox and Chrome depend on GConf.
By the way, Linux doesn't have anything like Active Directory because that's not how Linux (and UNIX before it) was designed to be deployed. If anyone tried to set up a modern Linux-based office, it would use thin clients and a central mainframe.
-
@boomzilla said:
@bjolling said:
We're mixing two discussions. The whole API topic was about applications using the correct built-in mechanism to avoid unnecessary rebooting.@bjolling said:
Yes, MSI is a fine thing, but how does it check for updates when they are available?Google seems very unhelpful atm to help me find back the API I used back in the NT days. I suppose that the relevant documentation on MSDN has been taken down. I'll post a link if I come across it.
duckduckgo.com has no trouble finding something. It was called the Windows NT Setup API. Which was being replaced with the "New Windows Installer" as it's called in this article
Do you agree that on Linux you are also just using a 3rd party package to keep your system up-to-date? On some distributions that application is APT that does all these things for you. I remember from years ago that on RedHat it was (is?) RPM. Someone managing some distro decided to create a repository and you only get updates if the repository manager approves of an update. If you have an application that is not approved/included by the distro manager, then you have nothing.
If someone would write an APT for Windows, they wouldn't have to start from scratch but just put a nice GUI and command line in front of the existing APIs. For example for keeping my Windows tools up-to-date I use http://chocolatey.org/. For development libraries I use http://www.nuget.org/. I even created my own development library on http://www.myget.org/. Why do you consider this inferior to using APT on Linux? It's all in the same spirit, just on a different platform.
Most of Windows bashing just comes from not knowing the platform.
-
@MiffTheFox said:
God forbid your distro goes obsolete (like Ubuntu after a year) and stops providing updates until you update the OS itself (what does this even mean in Linux terms, I've updated several core components (Linux, coreutils, X, Gnome) from APT but suddenly I need to upgrade "Ubuntu" and that's different?)
It basically just means you start pulling from different lists of available updates.
A distro upgrade changes your software sources then does a slightly modified sort of normal upgrade. (The modifications are to make it handle the dependency issues involved in what is commonly an update of every package on the system at once in a more sensible way than it would if that happened as part of a regular update.)
In "rolling release" distros, the software sources always stay the same, so there's no such thing as a distribution upgrade.
-
@MiffTheFox said:
Wow, I'd like to see what distro you use if it doesn't have a single hint of GNOME, and therefore, GConf being the one true way to store configuration. Must have been hard finding a popular browser, since both Firefox and Chrome depend on GConf.
Debian Squeeze as a server. No X installed, no GNOME installed, no KDE or GConf or . . .I have Ubuntu on my laptop. I can still edit the GConf config files with a text editor if I need to. Yes, they're XML, but they're still easier to deal with than the binary blob that is the Registry.
-
@bjolling said:
Not exaxtly. You can specify a lock when opening a file. If the coder decides the open and lock a file, it's not Microsofts fault. Notepad++ for example can open files without locking them.
I know this is way off-topic, but could someone please explain what's so appealing about Notepad++? Because I cannot stand the thing. It feels like someone tried to clone EditPad and got all the features almost right but subtly wrong just about every time. And how is anyone supposed to take a program seriously that uses a multitab interface but can't even get the fundamentals right, like closing a tab when the universal Close Tab shortcut, CTRL-F4, is pressed?
Why is it that this program is so popular when it's so awful?
-
@Mason Wheeler said:
I know this is way off-topic, but could someone please explain what's so appealing about Notepad++? Because I cannot stand the thing.
Whoa we agree.
@Mason Wheeler said:
And how is anyone supposed to take a program seriously that uses a multitab interface but can't even get the fundamentals right, like closing a tab when the universal Close Tab shortcut, CTRL-F4, is pressed?
Not to mention it can't handle overflow in menus correctly. You know, that thing that everybody else perfected in 1984? And that Microsoft provides a nice free API to take care of for you? Yeah they broke it somehow. Even that wouldn't be SO bad except it ships with a menu that's so long it's almost guaranteed to expose their error.
-
@Mason Wheeler said:
@bjolling said:
Not exaxtly. You can specify a lock when opening a file. If the coder decides the open and lock a file, it's not Microsofts fault. Notepad++ for example can open files without locking them.
I know this is way off-topic, but could someone please explain what's so appealing about Notepad++? Because I cannot stand the thing. It feels like someone tried to clone EditPad and got all the features almost right but subtly wrong just about every time. And how is anyone supposed to take a program seriously that uses a multitab interface but can't even get the fundamentals right, like closing a tab when the universal Close Tab shortcut, CTRL-F4, is pressed?
Why is it that this program is so popular when it's so awful?
Notepad++ is more full featured then Notepad2 and Scintilla is probably the best of the OSS editors/editor components for Windows?
Also the universal Close Tab shortcut is Ctrl+W.
-
@MiffTheFox said:
@Mason Wheeler said:
@bjolling said:
Not exaxtly. You can specify a lock when opening a file. If the coder decides the open and lock a file, it's not Microsofts fault. Notepad++ for example can open files without locking them.
I know this is way off-topic, but could someone please explain what's so appealing about Notepad++? Because I cannot stand the thing. It feels like someone tried to clone EditPad and got all the features almost right but subtly wrong just about every time. And how is anyone supposed to take a program seriously that uses a multitab interface but can't even get the fundamentals right, like closing a tab when the universal Close Tab shortcut, CTRL-F4, is pressed?
Why is it that this program is so popular when it's so awful?
Notepad++ is more full featured then Notepad2 and Scintilla is probably the best of the OSS editors/editor components for Windows?
Also the universal Close Tab shortcut is Ctrl+W.
In Windows, there are two universal Close Tab shortcuts, and the one that habitual keyboard shorcut users are most likely to use is CTRL-F4, because it's similar to ALT-F4, the universal Windows shortcut for Clsoe Window. If your program is running a multitab interface and does not support it, it's doing it wrong.
-
@bjolling said:
We're mixing two discussions. The whole API topic was about applications using the correct built-in mechanism to avoid unnecessary rebooting.
Yes. I brought up this topic of how Windows is so far behind, e.g., Debian when it comes to managing software updates as an aside.
@bjolling said:
Do you agree that on Linux you are also just using a 3rd party package to keep your system up-to-date? On some distributions that application is APT that does all these things for you.
There's nothing 3rd party about APT. It's how most of the system is installed to begin with.
@bjolling said:
If someone would write an APT for Windows, they wouldn't have to start from scratch but just put a nice GUI and command line in front of the existing APIs. For example for keeping my Windows tools up-to-date I use http://chocolatey.org/. For development libraries I use http://www.nuget.org/. I even created my own development library on http://www.myget.org/. Why do you consider this inferior to using APT on Linux? It's all in the same spirit, just on a different platform.
I don't think they need APT for Windows. MS already has a system developed to do all of this. They just need some sort of API to add additional places to look for updates. This should be provided by the operating system so that everyone can use this and developers never have to write another update checker for Windows
Having a third party package that can manage everything would be better than what you currently have (although I don't think you could include Windows Update in that tool). But it won't be deployed everywhere that Windows is so it will lose a lot of effectiveness.
@bjolling said:
Most of Windows bashing just comes from not knowing the platform.
Most anything bashing generally comes from ignorance. In this case, however, the defense of Microsoft comes from not knowing how other platforms have solved the problem more thoroughly than MS.
-
@Mason Wheeler said:
In Windows, there are two universal Close Tab shortcuts, and the one that habitual keyboard shorcut users are most likely to use is CTRL-F4, because it's similar to ALT-F4, the universal Windows shortcut for Clsoe Window. If your program is running a multitab interface and does not support it, it's doing it wrong.
Why would Ctrl+F4 be the "more likely" one? Ctrl and T are closer together, you don't have to leave the typing position, it's got platform synergy with ever other OS (Ctrl+W in GNOME/Linux and Command+W in OSX), and it's by analogy with Ctrl+T, or Open Tab.
Wikipedia lists the shortcut as Ctrl+W, although it does have Ctrl+F4 as a shortcut for Linux that's not supported everywhere.
Chrome:
Opera:
Internet Explorer:
Firefox lists both, but puts Ctrl+W primary. I do not have Firefox installed, nor do I ever want to again, so I cannot see what it shows in the browser itself.
True Ctrl+F4 is what the spec says but when even Microsoft themselves flout it (IE image I forgot to add and upload), it's time for a change.
-
"Congratulations, sir. We have completed all the necessary repairs to your automobile. Now if you would please turn off the engine and restart it, we will be finished."
Restarting the computer ensures that ALL caches and buffers are flushed, even caches that the application doesn't know about, even in-RAM caches that Microsoft installed last week.
As for running at start-up, yes, I agree with you. But that's a typical ego problem of developers. Developers naturally think that their application is the most important program in the world, and should take over 98% of the computer just to make it optimal. We have horror stories of applications mangling the system just to make that progam work. It's a generic problem, but bad marketing - think of it as evolution in action.
-
@ubersoldat said:
Is this a consecuence of the Windows Registry?
No, it's a consequence of Windows never having embraced the idea of an inode.
Every file in a typical Unix filesystem has an inode, a structure containing all the file's metadata (mod times, permissions etc). Directory entries are just names with a link to an inode, and each inode has a reference count to keep track of how many things link to it. This is the mechanism behind Unix's hard links, where you can make the same file appear in as many places in its filesystem's directory tree as you like.
When you delete a file from a typical Unix filesystem, what you're actually deleting is just a directory entry, not the file itself. Actual file removal doesn't happen until the inode's reference count drops to zero, which will only happen when nothing links to it any more.
When a Unix process opens a file, the resulting file handle contains a direct reference to the file's inode, and the inode's reference count is bumped by one. Effectively, the file handle becomes an extra, anonymous hard link to the file for the process's own use, and is completely unaffected by subsequent directory moves or deletions referring to the same inode. The inode's reference count stays nonzero, and the file stays available on the disk, for as long as any directory entry links to it or any process has an open file handle on it.
Windows filesystems, by contrast, have historically kept metadata in the directory entry itself. NTFS does have inode-like structures and the ability to create hard links, but the Win32 API came before NTFS and the standard file operations are all built to work with FAT filesystems that don't have that stuff. So when a Windows process opens a file, the directory entry it used to do that becomes locked and the system won't let you remove or rename it.
One consequence of this is that DLL files in use by any running process simply cannot be replaced. So whenever that needs to happen, as in often the case during software upgrade or removal, Windows does the usual thing: it works around a consequence of its own short-sighted design with another piece of short-sighted design. The replacement files get stuffed in some temporary directory somewhere, and their names get added to a Registry-based list of Things To Rename On Next Boot.
Boot-time renaming supports renaming new files over existing ones, in which case the existing ones get deleted; it also supports renaming files to nothing, which means the renamed files get deleted. It happens early enough that any files affected by it will probably not be open at the time. Still has gotchas, though.
This kind of thing is why I so prefer working with Unix and Unix-alikes over working with Windows. As must be the case for any complex system, Unix has many, many faults and flaws; but with a few notable exceptions, the ones that have given me grief look like consequences of deliberate and reasonable engineering tradeoffs and often lead me in the direction of finding simpler and/or more robust approaches to whatever problem revealed them. Unix and its derivatives still feel designed after all this time. Designed by sane and competent people, even. And I totally don't mind the fact that bits of the underlying machinery remain visible and retain a cavalier, almost feral disregard for operator safety.
I'm completely convinced that there has been far more design effort put into Windows, but mostly because there needed to be. When you choose spakfilla and high-gloss paint as primary construction materials, it takes a tremendous amount of skill to stop your tank from falling to bits on the highway.
-
@flabdablet said:
And I totally don't mind the fact that bits of the underlying machinery remain visible and retain a cavalier, almost feral disregard for operator safety.
... yeah.
-
I absolutely agree with your complaints about people with no sense of history. Makes me sad as well.
Not so sure I agree with you on the inherent superiority of the Registry over /etc. From my perspective the Registry just looks like a specialized loop-mounted filesystem optimized for small data, and it seems to me that just about any modern filesystem should be capable of implementing exactly the same functions natively; I'm not aware of any operation you can do on a registry key that isn't analogous to a directory operation in a general-purpose filesystem, or any operation you can do on a registry value that doesn't have a file analogue.
It also seems to me that if you squint a bit and look at /etc, what you see is a Registry full of (often quite long) REG_SZ and REG_MULTI_SZ values. I personally prefer dealing with these over dealing with the (often equally long) REG_BINARY blobs that litter the Windows registry (every tried force-disabling the keyboard Sleep button on a pre-Vista Windows box? It's ugly). One of the reasons why I prefer that is that config files in /etc frequently contain enough internal documentation to make them comprehensible.
I've always liked the fact that so much of Unix configuration is done with text files. Yes, it does impose a parsing load. Yes, there are lots of different parsers. Yes, each application program has its own config learning curve. No, vi is not a stellar configuration management GUI. But for lots of small admin tasks it's perfectly adequate. For larger tasks, especially scripted ones, being able to use a small and reasonably well standardized set of text processing tools on configuration stuff, without needing an intermediate translation tool like REG, is very useful. So is being able to manage all of /etc with a source code control system.
Every complex system will have a bunch of stuff that's baked in, and a bunch of stuff that's essentially just convention. I'm generally irritated a lot less often by the conventions that have accreted around Unix than those around Windows. I'm offended a lot less frequently by design aspects of Unix than of Windows, which generally strikes me as having a too much, too late approach.
But for the sake of everybody's sanity and self esteem, do you think we could all agree that it's good to have a vast range of tools available, it's good to be able to use the tools one prefers and feels comfortable with, and that tool preferences do not define one's intelligence, or competence as a programmer or administrator, or worth as a human being? This "$system (or should that be %system%?) is a piece of shit and you're a dumbshit if you don't agree" bickering is all pretty tiresome.
-
@blakeyrat said:
@Lorne Kates said:
(Though I still stand by my original point: Microsoft force installing Skype on home PC via Windows Update is a bad idea for everyone. It's bad PR for Microsoft. Users with violated trust will shut off automatic Windows Update, thus its bad for them due to lowered security. Skype gets a bad reputation for being malware. No one wins).
But users don't get it. That's the point you're still not getting.
Maybe the 0.00001% of users who do retarded things or have retarded network administrators got it. But "users" as a generalization were not affected in the slightest.
I've made absolutely no changes to the out-of-the-box Windows Update settings, on neither my nor my wife's computer. And yet, poof, Skype. So pray tell, Master of Windows, what non-retarded setting should I have "chosen", if "the default security settings Microsoft provided" aren't the right ones? And before you answer, I will gladly sift through other threads where you lambasted people (ie: me) for daring to questing Microsoft's default settings.
-
@boomzilla said:
@bjolling said:
I fundamentally disagree. APT is not part of Linux. My RedHat installation doesn't have it. It uses RPM. A quick search on Google revealed that people are actually posting on how to get APT working for RedHat. For Mandrake, they are talking about 'urpmi' .Do you agree that on Linux you are also just using a 3rd party package to keep your system up-to-date? On some distributions that application is APT that does all these things for you.
There's nothing 3rd party about APT. It's how most of the system is installed to begin with.So which one of the above the the linux package manager?
-
@blakeyrat said:
@ASheridan said:
So if an INI file goes beyond 32KB it's no longer an INI file? Really? Seems Microsoft thinks it can still be called an INI above 64KB, so stop being so fucking stupid and pedantic.I found a reference to the limit, which appears to be some weird-ass Windows-specific limit. So tell me again, why can't an INI file be greater than 32KB?
THE ENTIRE FUCKING FORMAT was created by Microsoft for use in Windows 1, 2, and 3. THAT IS WHY THE FORMAT EXISTS. It's not a weird-ass "Windows-specific limit", it's DEFINED IN THE FILE FORMAT. THAT IS THE WHOLE POINT OF THE FORMAT.
Jesus fuck does nobody have ANY sense of history in the computer industry at all?
-
@blakeyrat said:
@boomzilla said:
I don't recall him saying a scheduler sucked, only that it sucks on Windows that there's no central place to update apps, each app that isn't Microsoft has to keep itself updated.I'm trying to follow you here, but now you seem to be denying the existence of Windows Updates (or whatever Lorne wants it to be called). Have you really never used Windows?
Windows Update runs on a schedule. Which is (according to you) something that "sucks".
It's no fault of the app developers. Pick apps that are commong between Windows and Linux: OpenOffice, Firefox, Chrome, and you'll see that they all have different ways of keeping themselves updated on Windows, but in Linux they all just use whatever package manager is used by that distro.
-
@Mcoder said:
But the great thing is that once you add those 3rd party repositories, they are then managed by the package manager, which informs you of available updates. This is what Windows needs.@MiffTheFox said:
The problem with APT or any Linux package manager is that you're limited to whatever whims a particular distro has about what software to put on there.
Well, if you don't want to add third party sources, yes, you are restricted to the distro.
-
@bjolling said:
I fundamentally disagree. APT is not part of Linux. My RedHat installation doesn't have it. It uses RPM. A quick search on Google revealed that people are actually posting on how to get APT working for RedHat. For Mandrake, they are talking about 'urpmi' .
So which one of the above the the linux package manager?
RedHat uses yum, Debian uses APT. And the nature of Linux means that you can alter your system in very low level, fundamental ways if you're motivated enough. So what?
I know that Windows guys hate the whole distro vs "Linux is just a kernel thing", but when you're comparing an installation of Linux vs Windows, the distro (e.g., RedHat, Debian, Slackware) is pretty important. I guess it's important to understand that words can mean different things when used in different ways. And you really can't separate the package manager from the rest of the OS. Well, maybe if you built everything from sources by hand or something (but I think we can safely ignore that sort of lunacy for this thread).
But each distro comes with a package manager. And I think any non-pedantic dickweed definition of the operating system has to include that component.
-
@boomzilla said:
I know that Windows guys hate the whole distro vs "Linux is just a kernel thing", but when you're comparing an installation of Linux vs Windows, the distro (e.g., RedHat, Debian, Slackware) is pretty important. I guess it's important to understand that words can mean different things when used in different ways
I suppose you meant this as a remark about Windows guys in general, but for me personally the Linux way helped me in realizing that Microsoft not only delivers the core OS but also a distro. Albeit the single official distro.But users can already choose their preferred Windows shell: cmd.exe, powershell, explorer.exe, cygwin any many others. With Windows 8 their will be 2 Window managers included: explorer desktop style and metro style... (http://www.zdnet.com/blog/microsoft/heres-the-one-microsoft-windows-8-slide-that-everyone-wants-to-redo/10736). I can imagine that someday somone will put together a distro on top of the Windows 8 Core Services... http://en.wikipedia.org/wiki/MinWin (http://i.zdnet.com/blogs/minwin.png?tag=content;siu-container)
I guess that until now the official Windows distro has been configurable enough for most people so nobody feels the needs to create a separate one.
-
@bjolling said:
@boomzilla said:
I know that Windows guys hate the whole distro vs "Linux is just a kernel thing"...
I suppose you meant this as a remark about Windows guys in general, but for me personally the Linux way helped me in realizing that Microsoft not only delivers the core OS but also a distro. Albeit the single official distro.It was more of a pre-emptive strike against a certain breed of pedantic dickweedery.
@bjolling said:
I guess that until now the official Windows distro has been configurable enough for most people so nobody feels the needs to create a separate one.
Which "official" distro? Server? Home? Ultimate? They just added the RT thing. One thing that has annoyed me for quite a while is that they disable using remote desktop to connect to the non-enterprise versions. Yeah, because no one has a home network and wants to check in on, say, their kids' machines. Grr. So...configurable for a price.
Anyways, I don't understand why you think people might not want to. They're not really allowed to, since, uh, it's all owned by Microsoft. I guess hardware vendors can implement a certain amount of customization on the OEM copies they sell. Either way, this all supports my point about the package managers and their relationship to the OS.
-
@boomzilla said:
@blakeyrat said:
I know you're Boomzilla and thus functionally retarded (I fell sorry for you), but how does that differ from Linux?
Haha, you found a typo. I guess you were too excited by that to notice that I was simply using bjollings ignorant trolling to respond to him (the typo was mine)?
@blakeyrat said:
@boomzilla said:
TRWTF is a modern OS that requires 3rd party software to use their own WTF-y update checkers in order to keep software updated in a non-manual fashion.
Because Windows requires that? Or... are you just plopping down random words here and hoping we won't notice that you have aphasia?Yes, Windows does require that, AFAIK. Do you know which part of Windows will check for and install third party software updates? And when you find it, please let Adobe, et al know, too, so they can stop deploying their WTF-y update checkers. I've mentioned this several times around here. Including a discussion about this on your thread where you demonstrated your ignorance of how all this works on a Linux distro that uses APT.
The point here is you still have to (AFAIK) check in with your own repository, but you don't need to leave a program running all the time like Adobe and Oracle do: instead, you set it up to run with Scheduled Tasks. Once a day, once a week, once a month, whatever, your program will get called. It starts up, checks your server for an update, offers it to you if necessary, and then shuts down, instead of running continuously in the background. This is what Chrome does. Of course, Chrome takes it one step further and updates itself for you, behind your back. If you're the kind of person who has a few tabs open but shuts the browser down when done with it, you'll silently get the new version the next time you run it. Jeff Atwood calls this the infinite version number. If you're the kind of person that never closes the 50 tabs he keeps open, eventually the browser will say "hey, fella, how 'bout lettin' me restart?"
-
@FrostCat said:
@boomzilla said:
Yes, Windows does require that, AFAIK. Do you know which part of Windows will check for and install third party software updates? And when you find it, please let Adobe, et al know, too, so they can stop deploying their WTF-y update checkers. I've mentioned this several times around here. Including a discussion about this on your thread where you demonstrated your ignorance of how all this works on a Linux distro that uses APT.
The point here is you still have to (AFAIK) check in with your own repository, but you don't need to leave a program running all the time like Adobe and Oracle do: instead, you set it up to run with Scheduled Tasks. Once a day, once a week, once a month, whatever, your program will get called. It starts up, checks your server for an update, offers it to you if necessary, and then shuts down, instead of running continuously in the background. This is what Chrome does. Of course, Chrome takes it one step further and updates itself for you, behind your back. If you're the kind of person who has a few tabs open but shuts the browser down when done with it, you'll silently get the new version the next time you run it. Jeff Atwood calls this the infinite version number. If you're the kind of person that never closes the 50 tabs he keeps open, eventually the browser will say "hey, fella, how 'bout lettin' me restart?"So, you have been comprehending this discussion about as much as blakeyrat, eh?
-
Ravenous cheese trees.
-
-
@Lorne Kates said:
I'm not using WSUS. I'm using Windows Automatic Updates, which is also known as WSUS.
No, it's WUAU, Windows Update Automatic Updates: The service is named wuauserv, now wsusserv. Ancillary files in your %system32% are wuauclt.exe and wuaueng.dll (maybe the latter will have a different name if you're not using English.)
If anyone you know is calling WUAU WSUS, you should tell them to stop.
-
@FrostCat said:
@Lorne Kates said:
I'm not using WSUS. I'm using Windows Automatic Updates, which is also known as WSUS.
No, it's WUAU, Windows Update Automatic Updates: The service is named wuauserv, now wsusserv. Ancillary files in your %system32% are wuauclt.exe and wuaueng.dll (maybe the latter will have a different name if you're not using English.)
If anyone you know is calling WUAU WSUS, you should tell them to stop.
WSUS is a centralized update server which is usually used in corporations so they can download a windows update once, test it, and then roll it out to their users once they are satisfied that it won't have a negative effect on the users (i.e. interfere with some home grown or legacy app). It also prevents the same update from being downloaded hundreds of times from microsoft which saves bandwidth.
http://technet.microsoft.com/en-us/windowsserver/bb332157.aspx
It works in conjunction with WUAU - you point your WUAU to the WSUS instead of the MS repositories.
-
@flabdablet said:
Windows filesystems, by contrast, have historically kept metadata in the directory entry itself. NTFS does have inode-like structures and the ability to create hard links, but the Win32 API came before NTFS and the standard file operations are all built to work with FAT filesystems that don't have that stuff. So when a Windows process opens a file, the directory entry it used to do that becomes locked and the system won't let you remove or rename it.
One consequence of this is that DLL files in use by any running process simply cannot be replaced. So whenever that needs to happen, as in often the case during software upgrade or removal, Windows does the usual thing: it works around a consequence of its own short-sighted design with another piece of short-sighted design. The replacement files get stuffed in some temporary directory somewhere, and their names get added to a Registry-based list of Things To Rename On Next Boot.
Actually, strictly speaking, neither of these are true; CreateFile can be opened with FILE_SHARE_DELETE, and can be demonstrated with a 6-line or so program. Also, and i'm not saying this wouldn't be more work, but if you dynamically load your DLLs, you can unload them as well--then it would be a matter of having a protocol such that your app's updater can tell you to release handles and when it's safe to reload them. Sure, nobody does this, but it's certainly possible. It's not trivial to demonstrate so I'm not going to bother to write the program.
But on one level, we're straining our metaphors. In the real world, you wouldn't replace your car's spark plugs while your car's running, and asking to keep an application running while it's being updated is similar.
-
@boomzilla said:
@bjolling said:
I fundamentally disagree. APT is not part of Linux. My RedHat installation doesn't have it. It uses RPM. A quick search on Google revealed that people are actually posting on how to get APT working for RedHat. For Mandrake, they are talking about 'urpmi' .
So which one of the above the the linux package manager?
RedHat uses yum, Debian uses APT. And the nature of Linux means that you can alter your system in very low level, fundamental ways if you're motivated enough. So what?
I know that Windows guys hate the whole distro vs "Linux is just a kernel thing", but when you're comparing an installation of Linux vs Windows, the distro (e.g., RedHat, Debian, Slackware) is pretty important. I guess it's important to understand that words can mean different things when used in different ways. And you really can't separate the package manager from the rest of the OS. Well, maybe if you built everything from sources by hand or something (but I think we can safely ignore that sort of lunacy for this thread).
But each distro comes with a package manager. And I think any non-pedantic dickweed definition of the operating system has to include that component.
Linux means Ubuntu*, obviously. Outside the offical repository, where can I find a MySql repository for Pacman? Or an OpenOffice one for Portage?
* Technically, it means Debian and it's descendants, and maybe RedHat's descendants too.
-
I know what WSUS is, but that's not the point. What you're apparently inferring is that WSUS isn't WUAU, right? I mean, what you said is true but not relevant to my post.
ETA: Oops, quote failure. I'm replying to rad131304's reply to me, explaining what WSUS is to, I guess, people who don't know.
-
@boomzilla said:
I know that Windows guys hate the whole distro vs "Linux is just a kernel thing", but when you're comparing an installation of Linux vs Windows, the distro (e.g., RedHat, Debian, Slackware) is pretty important.
Absolutely true, which is why in some of my posts I took to saying "Linux distribution" instead of just Linux.Some time ago, one of our UNIX guys came to me complaining about how Linux is "all supposed to be the same", but that different distributions put files in different places. I told him that it's akin to saying you administer UNIX, but it depends on what implementation. HP-UX, Solaris, SCO OpenServer, etc. all do things differently and put thing in different places, yet they are all UNIX. The only commonality between all Linux distributions is simply that they use the Linux kernel as a base, nothing more.
-
@MiffTheFox said:
@boomzilla said:
But each distro comes with a package manager. And I think any non-pedantic dickweed definition of the operating system has to include that component.
Linux means Ubuntu*, obviously. Outside the offical repository, where can I find a MySql repository for Pacman? Or an OpenOffice one for Portage?
* Technically, it means Debian and it's descendants, and maybe RedHat's descendants too.
I'm nearly certain that you thought you had a point. Would you mind sharing it for us?
I'm not sure what you mean by "official repository," and why you'd want to find mysql in that other repository. I don't know if anyone other than Arch Linux uses pacman, but you can get mysql here. Hmm, now I'm a little less certain about your thoughts about your alleged point.
-
@boomzilla said:
Anyways, I don't understand why you think people might not want to. They're not really allowed to, since, uh, it's all owned by Microsoft
I never heard that one before so I would be interested if you could link to some "terms of use" by Microsoft that tries to enforce it. AFAIK the only reason I can't write a window manager for Windows is because I'm not familiar enough with how the Windows Core components work. I also don't care to investigate because I'm more than happy with the existing ones. If you had said: "People can't write a window manager because the core components are closed source and documentation is hard to find," I probably would have agreed.@boomzilla said:
I guess hardware vendors can implement a certain amount of customization on the OEM copies they sell. Either way, this all supports my point about the package managers and their relationship to the OS
I don't see how it does but since we have both articulated our point of view, I'm going to leave it at that if you don't mind.
-
@boomzilla said:
I'm not sure what you mean by "official repository," and why you'd want to find mysql in that other repository. I don't know if anyone other than Arch Linux uses pacman, but you can get mysql here. Hmm, now I'm a little less certain about your thoughts about your alleged point.I was using OpenOfice as an example because it's what ASheridan was using. As for MySQL, note that Arch only has 5.5. If I wanted 5.6 I'd have to install a binary tar or compile it by hand. The odds of finding a useful package go down the more obscure the software is.
-
@MiffTheFox said:
@boomzilla said:
I'm not sure what you mean by "official repository," and why you'd want to find mysql in that other repository. I don't know if anyone other than Arch Linux uses pacman, but you can get mysql here. Hmm, now I'm a little less certain about your thoughts about your alleged point.I was using OpenOfice as an example because it's what ASheridan was using. As for MySQL, note that Arch only has 5.5. If I wanted 5.6 I'd have to install a binary tar or compile it by hand. The odds of finding a useful package go down the more obscure the software is.
Uh, OK. I'll assume you had a good reason to share that epiphany with us. 5.5 seems like the current release version, so I'm not sure what your point has morphed into. Arch, being a rolling-release distro, tends to have recent stuff. Of course, release/version/packaging policy is a big discriminator among distros.
-
-
I can't fuck around with reading this massive thread, but my CtrlF didn't find mention of PendMoves.
http://technet.microsoft.com/en-us/sysinternals/bb897556.aspx
Lots of time these shit-apps tell you to restart for no reason.
