Windows 7 Lets You Uninstall IE8 474
CWmike writes "A just-leaked build of Windows 7 lets users remove Internet Explorer, the first time that Microsoft has offered the option since it integrated the browser with Windows in 1997, two bloggers reported today. The move might have been prompted by recent charges by the European Union that Microsoft has stifled browser competition by bundling IE with its operating system, the bloggers speculated. One solution under consideration by the EU would require Microsoft to disable IE if the user decided to install a different browser, such as Mozilla's Firefox or Google's Chrome. Microsoft had no comment when asked to confirm whether Windows 7 will let users dump IE8 or whether the option was in reaction to the EU charges."
Re:what about accessing windowsupdate via browser? (Score:5, Informative)
Re:You can already do this ... (Score:5, Informative)
Read the article. They state that iexplore.exe is gone.
Sure, some libraries will stick around. They have to, otherwise a lot of applications will break. You can't "decouple" a dependency from applications without breaking them. But IE was never integrated into the kernel; it was integrated into the shell. I know that doesn't jive with your particular interpretation of the definition of an "operating system", but that is the reality of the situation.
Re:You can already do this ... (Score:5, Informative)
The kernel isn't the operating system. That's the basis of the GNU/Linux vs. Linux debate.
That said, this seems to be functionally comparable to deleting the Safari.app on a Mac - the application is gone and cannot be launched, but the rendering engine sticks around because it's used elsewhere in the operating system for other tasks.
Re:what about accessing windowsupdate via browser? (Score:4, Informative)
At least be informed in your trolling.
Re:Uninstall? Yeah, right... (Score:2, Informative)
New Prank (Score:5, Informative)
Looks like the newest prank to play on someone's computer will include uninstalling all of their browsers.
Re:Windows updates? (Score:4, Informative)
Re:Uninstall? Yeah, right... (Score:2, Informative)
How do you uninstall a program and all the dependencies it installs in Linux?
Not being an ass - i'm just genuinely curious. I've never found a way easier than windows.
apt-get purge program
apt-get autoremove
That should work for apt-based distros.
Re:Uninstall? Yeah, right... (Score:3, Informative)
and the application data folder?
what about the localstore?
did it place any files in %windir% or %sysdir%?
did it make any file extension associations?
did it add any environment variables?
etc.
crap cleaner won't clean -all- of that up.
That said, the original poster's comment was bunk; an uninstallation isonly as good as the uninstall routine. If it doesn't delete -all- files / remove -all- registry entries, etc. set upon install, then that's an issue with the uninstaller, not with the host OS.
I'm sure that some of the -package managers- do a great job at tracking this (though they're likely to miss run-time file/store changes just as well), but that says far more about the package manager than it does about the host OS.
Your best bet is going to be to take a snapshot of your system, install, run for a while, do a diff, remove known variables from other use (from earlier diffs, presumably) - i.e. e-mail database, temporary files, etc. - store that and use that to remove files/registry settings/etc. later on.
Re:Uninstall? Yeah, right... (Score:3, Informative)
Not quite; a number of apps put stuff outside of the .app wrapper directory. Anything that loads a kernel extension (vmware, for example), as well as other application that put frameworks in /Library and /System/Library. And then there's prefs and cache files left over in your own Library directory.
Still, it's significantly better than it is on windows.
Re:At last! (Score:4, Informative)
I suspect it would fail when attempting to delete the deltree binary itself, or the directory it belongs to. Haven't tried, though.
(No such problem on Linux, of course; rm -rf / will happily wipe your entire fs, including the rm binary and the /bin directory.)
Re:Disable IE? (Score:1, Informative)
I will be outraged if the default behavior is "disable IE on installation of any other browser".
Nowhere in the article is such an automatic disabling of IE mentioned. You have too much imagination, and apparently get carried over by it too easily.
Re:You can already do this ... (Score:5, Informative)
Why couldn't this slashdot post point to the two people who actually came up with this? CWMike provided no original insight whatsoever.
Original sites referenced by CW's article:
http://www.aeroxp.org/2009/03/ie8-functionally-removable/ [aeroxp.org]
http://chris123nt.com/2009/03/03/win7-build-7048-ie8-is-removable/ [chris123nt.com]
rm -rf / (Score:4, Informative)
(No such problem on Linux, of course; rm -rf / will happily wipe your entire fs, including the rm binary and the /bin directory.)
This is a "bug". Under recent POSIX revisions this is now considered incorrect behaviour (something about trying to follow "/." and "/.."):
http://blogs.sun.com/jbeck/entry/rm_rf_protection
Supposedly Debian (from Sid onwards) also does not allow 'rm -rf /'.
Re:Why remove it alltogether? (Score:1, Informative)
That won't work. You need a shim layer. However, one is available.
Ship Wine's mshtml.dll (I think it interfaces to gecko rendering engine) and link to that. Wine's mshtml.dll is LGPL so your app is safe.
Re:Uninstall? Yeah, right... (Score:3, Informative)
Depends on the application/installer... but you should also check:
?:\Documents and Settings\%Username%\*
?:\Documents and Settings\%Username%\Application Data\*
?:\Documents and Settings\%Username%\Local Settings\Application Data\* (Hidden)
?:\Documents and Settings\%Username%\Local Settings\Temp\* (Hidden)
?:\Documents and Settings\%Username%\SendTo\* (Hidden)
?:\Documents and Settings\%Username%\Templates\* (Hidden)
Sometimes:
?:\Documents and Settings\All Users\
As well as:
?:\Program Files\Common Files\*
?:\Program Files\InstallShield Installation Information\* (Hidden)
?:\WINDOWS\Downloaded Installations\*
?:\WINDOWS\Inf\* (Hidden)
?:\WINDOWS\Installer\* (Hidden)
?:\WINDOWS\System32\*
?:\WINDOWS\Temp\*
?:\WINDOWS\dllcache\* (Hidden)
?:\WINDOWS\Drivers\*
And the registry:
HKCR\Software\%Document Types%
HKCU\Software\CLSD\*
HKCU\Software\%App/DevName%
HKCU\Software\Microsoft\Windows\CurrentVersion\Explorer\MenuOrder\*
HKLM\Software\%App/DevName%
HKLM\Software\Microsoft\Windows\CurrentVersion\Explorer\MenuOrder\*
HKLM\SOFTWARE\Microsoft\Windows\CurrentVersion\Installer\*
HKLM\SOFTWARE\Microsoft\Windows\CurrentVersion\SharedDlls\*
And quite a few others, I generally just search for DeveloperName, delete (most) matches, then search for ApplicationName, delete matches, ExecutableName, delete matches...
Re:At last! (Score:2, Informative)
Actually it doesn't. You have to unmount all your pseudo filesystems first and it's been a long time since Linux let you do that. I tried it on several systems when I was working at Turbolinux and regularly wiping boxes for "clean" installation tests.
Vintage 4.2 BSD could do it (I watched someone do it on purpose, he was wiping the system for a reinstall). System V-oid boxes could not, due to lacking an rmdir(2) system call and forcing an implementation of rm -r by doing a fork-exec /bin/rmdir (which was setuid root). My first intentional /bin/rm -rf / ended with an endless sequence of `/bin/rmdir: not found' messages.
Re:I don't understand what is so complicated (Score:4, Informative)
That was the case up until IE 7, but I things changed more there. And iexplore.exe and explorer.exe have always had separate process spaces, even back in the IE6 days. (Very important since an explorer.exe browser crashing requires restarting all of explorer.exe including the desktop. That gets very annoying, so using IE processes to browse the web have always been a good idea.)
And technically, IE6 was still a seperate program that just ran the same code as explorer.exe did for both browser and file modes.
When IE 7 is installed the explorer.exe is prevented from entering a web-browser mode. While the trident engine remains in the Windows core, the UI engine that IE7 uses is completely different. With IE7 and an old copy of IE6's iexplore.exe one can load the old IE 6 UI, but the IE 7 version of the trident engine is still used. The fact that the iexplore.exe of ie 6 can use browser mode, but explorer.exe cannot after the installation of IE7 does show that iexplore.exe has always been more than a stub.
Re:Disable IE? (Score:1, Informative)
This is not, nor has it ever been, about end users.
If HP includes Firefox on their machines, having the IE binary around means that HP has to support two browsers, because that's what they shipped. If HP cannot remove the IE binary in any MS-supported fashion (which they can't right now), then the support cost for Firefox is roughly double the support cost for IE. The only way to cut support costs is to ship only with IE, even if it's not the better product by any other measure.
Microsoft did the exact same thing to web developers. IE used to support Netscape-style plugins AND ActiveX controls. So cash-strapped web developers made Netscape-style plugins because they'd work on more browsers. But once they had majority marketshare, Microsoft pulled support for Netscape-style plugins. This meant that developing plugins for multiple browsers suddenly cost twice as much as it did before. Cost-conscious developers picked one browser and developed for it. Since Microsoft waited until they had majority marketshare to do this, which one did they pick? Easy.
Tons of choices are made before you, the consumer, ever buy your computer. Those decisions are often made for reasons of economic expedience. And although Microsoft rarely makes their products cheaper, they frequently succeed in making their competition more expensive.
The choices Microsoft has not yet figured out a way to prevent you from making are not what we're talking about.
Re:what about accessing windowsupdate via browser? (Score:5, Informative)
Server core still has IE libraries - for instance, WinInet which basically is a standard internet connectivity library is there. Hell, even Hyper-V server (the OS that is free and can only run Hyper-V) will actually get offered some IE updates - because some IE components are still part of the OS. Iexplore.exe isn't there, but other chunks are there because substantial parts of the OS (and even third-party applications) use them.
Re:At last! (Score:5, Informative)
So to have more protection over temporary data in a program, open a file and then immediately unlink it - only programs that can manage to open it between(/at) its creation and unlinking from the directory structure will be able to access the data within it; this also leads to situations where the total space allocated on a disk [partition] (looking at, say df) can be much larger than is obviously apparent (using, say du) - this can happen if you have a large log file that is being written and you rm the directory entry for it: only when the program filling the log exits will the space be released.
Re:rm -rf / (Score:4, Informative)
This is not Debian-specific. Just RTFM of rm(1) from GNU Coreutils and you'll see the option --preserve-root is enabled by default. To override it use --no-preserve-root. Mine's coreutils-6.12 here.
Of course you can see this as another disadvantage of GNU.
Re:I'm sure the EU will go after Apple too. Yeah. (Score:2, Informative)
there is, however, a certain monopolized desktop market that MS is using to expose their browser to everyone
Re:Why remove it alltogether? (Score:2, Informative)
Opera: 94.16% A+ Pass
Firefox: 60.40% C Pass
Safari: 64.23% C Pass
IE: 0.00% FAIL