I mean, I understand why the end-user interface should be integrated, or at least well linked together.
But the whole Unix philosophy, the whole reason it works so well, is that we tend to isolate things as "do one thing and well". If I were to try to compete with Exchange, I'd use Postfix for the actual SMTP server, probably something like RoundCube for the webmail, etc. Indeed, even Postfix can be broken down into tiny, replaceable programs that communicate via pipes.
Why would I want one monolithic app that handles everything?
Anyone who thinkos otherwise hasn't spent any/enough time in a real, productive office workplace.
I have, but it was a small office.
Let me ask, then: What, specifically, is missing from Google Apps in terms of calendar/email?
Google Apps, THANK GOD, doesn't deliver. Moving from platform lock in on the desktop to another platform lock in where not only the software that I'm using but also my frakkin' USER DATA is also locked in is literally jumping out of the frying pan and into the fire.
Except that your user data is mostly available, and in open standard formats. I don't know about Docs, but the Calendars can be downloaded as iCal, and the mail can downloaded via IMAP. And in the mean time, you've got a much nicer web interface that works cross-browser, pretty much by definition cross-OS, while Exchange pretty much works with Outlook.
Can't you idiots see that Google has the makings of an evil monopoly that makes Microsoft look like your friendly local corner store.
I can, but there are several big reasons I'm not nearly as worried about Google as I am about, say, Apple (who truly does make Microsoft look like a friendly corner store, at least on mobile devices):
First, Google stuff is technically superior to Microsoft stuff in most ways that matter to me. Chrome is a better browser than IE, and before Chrome, Google sponsored Firefox.
Second, Google does seem to have a real commitment to open source and open standards. See: Summer of Code, GTalk (uses Jabber), OpenID, ODF in Google Docs, Firefox, Chrome (the browser, the OS, and 99.999% of the support needed to build it, excluding some codec support), Android, native Linux ports of most of their apps (even proprietary ones), free APIs to their proprietary services (like maps -- Microsoft wants you to call them before they'll even give you a price), Gears, O3D, Protocol Buffers...
Microsoft does some interesting open source things, but they tend to prefer their own license to more standard, proven licenses such as BSD or GPL, and they also tend to do it half-assedly -- for instance, while Google will give away a VM (as with Dalvik), Microsoft will not give the source to the.NET runtime. Since so much of what they do lately (including open source stuff) depends on.NET, this is a major roadblock. Sure, they help Mono along, but for how long?
Perhaps I'm not being fair, but Microsoft has a long history of embrace, extend, extinguish. Google, on the other hand, has a history of create/embrace, extend, set free.
Not that it would particularly matter to me -- Chrome is pretty accurate, and it's actually open source and easily extendable (in the developer version).
Regarding memory: First, in these tests at least, the measurements of Chrome were wrong. And second, does it matter? I have 4 gigs of RAM. It's hard to remember the last time I went into swap, and I certainly can't remember it being because of a web browser.
Regarding crashing: I don't know about you, but I'd rather have my whole browser not crash than have it be able to recover my tabs when it does. Chrome does keep a "last closed" set of tabs, but I can also comfortably run a development release -- seriously a nightly build -- as my primary browser, and I haven't seen the whole thing crash yet. I've only been able to make tabs crash with the developer tools and/or my own badly written extensions -- not that it's an excuse, just pointing out, tabs don't crash a lot.
That's more than I can say for Firefox -- Firefox (stable) crashes less often than these Chrome tabs (again, development release, lots of hacking around on my part), but I've seen the whole browser go down, taking all my tabs with it.
Comes with the distribution repositories and is stable, maintainable and patched
Not yet, but working on it. In the meantime, Google has its own repositories.
Has effective script control (white-listing, base-domain)
Some of this could be done with a proxy.
But seriously, I have no idea how this is going to work, if they support it at all.
Has effective ad blocking capability
I wrote an adblocker for Chrome in two days. Sure, it was for the development version of Chromium, but you get the idea -- once the extension API is released, this is just going to be a non-issue.
Also, Iron (a fork of Chromium) has an adblocker built in. I'd be surprised if they don't have some sort of script control, also.
Does surf the web and performs adequately on my system
A good tester should bookmark about 200 sites in various categories, save passwords for about 20-30 sites, have some forms saved,
How do you do that scientifically? Even eight tabs is going to raise questions of fairness -- maybe one of those pages lagged during the test?
And, you're talking about some tech journalist writing an interesting article. Where are you going to find someone willing to actually do the test you're describing? Sounds like a lot of work.
We don't care whether the browser hogs half a gig or more.
I have four gigs of RAM. What else am I using it for?
And of course, see Hurricane78 for an explanation of why that's "hogging" half a gig of RAM. Keep in mind that these are sort of theoretical maximums -- anything the browser does for you above and beyond just rendering a page is going to want its own RAM, too.
We don't care whether it can render a page correctly or makes CSS look like a 5 year old had a field day with some sharpies.
Do any of the browsers in question look like a 5 year old with sharpies?
Chrome was the overall winner, and it uses Webkit, which has historically been pretty good with web standards.
crappy servers crammed into farms that oversold their capacity hundredfold and ISPs doing the same.
So you're saying, why not slow it down even more?
I have fiber. Some sites are going to be slow, but some sites will be fast. On the fast sites, the difference between a page loading instantly, or some script executing before I can blink, is nice compared to "a few seconds".
Did he, by any chance, disable the phishing/malware protection?
And did he actually open up a packet sniffer and see those URLs go to Google? Or is it possible this was a coincidence, seeing as it's only really happened once?
Firebug is built-in as the Chrome Development Tools.
Adblock is available in a fork of Chrome, called Iron. It's also trivial to create an adblocking extension to Chrome itself -- and someone's done this, it's just not maintained.
Firefox addons run in the same process as Firefox, likely in the same thread. Firefox tabs are similar. All it takes is one slow extension to slow down the entire experience.
Chrome, on the other hand, is implementing addons as just privileged webpages. This means that, except for the very small part of an addon that might be interacting with the current page, the addon won't block the browser -- it's mostly going to be running in a separate process. And even the content script that's running on the current page, well, there's one of those running per tab, so an extension being slow in one tab won't block another tab.
Not to mention, if you're going to implement a nice, cross-platform Firefox addon, you're doing it in Javascript/XUL. Chrome addons are Javascript/HTML. Thus, Chrome's faster Javascript engine does count here.
Your ??? is "install foo", where foo is the name of whatever you wanted.
Never mind the point-and-click GUI that is add/remove programs.
most will be able to read and understand the words that appear in the folder telling them to drag the icon onto the big Applications alias icon sitting beside it.
Of the Mac users I've seen, I've only seen two ends of the spectrum.
One is highly technical. They use a Mac because they like Unix, but they also like stuff Just Working, and they like the slick, shiny interface, and all the proprietary software available. These people would be perfectly capable of installing software from MacPorts or RubyGems if they had to, and indeed, Macs are popular among the Rails community.
The other is highly non-technical. They like a Mac because it's what they've always used, and they have no idea what goes on under the hood. It's these people who tend to ignore the instructions and focus on the pictures.
And it's here that Linux is actually easier -- because if someone includes a link to an apt-url, or to a deb, they can point and click, and it'll work, and there is no way they can screw it up and have it be half-installed as a disk image somewhere.
if your installer is in a Zip file, you open it and run it.
Unless the zipfile is of the program itself, in which case, you make a folder, drag the contents of the zipfile to that folder, and run it. And technically, you probably should be putting that folder somewhere in Program Files, and making it owned by whatever usually owns Program Files, and, you know, all the stuff that an installer normally does for you.
it's still corporate. As is Fedora and Red Hat and Debian.
You may well have a point with Ubuntu, but which corporation runs Debian?
When I mention "driver" to someone and they give me a funny look and say "What's a driver?", do you really think they will have an easy time of using Linux?
I suspect they already do, with Google and tons of websites they use, and possibly with their phone. As for the desktop, absolutely, yes.
Perhaps they will have a difficult time installing Linux, but they would very likely have an easy time of using it.
Put another way: Is that the kind of person who generally installs software on their computer? Is it the kind of person you want installing software on their computer?
you'll spend 20 minutes just trying to get them to open a terminal so you can give them the commands to fix whatever issues they have that cannot be fixed via the GUI.
After which I'll spend ten seconds typing those commands into an IM window, and having them copying and pasting them into that terminal.
And that assumes you're right about those 20 minutes. Most people I know can handle "Click the word Applications in the upper left of the screen. Then click Accessories. Then click Terminal."
But again, there's still that large assumption that such an issue will come up at all. In my experience, issues come up for people on Windows that require such calls much more frequently than on Linux.
What you think someone should or should not be allowed to do is irrelavent and imposing that on others frankly goes against the whole Linux and free software philosophy.
I beg to differ. If that was really the case, Linux would likely be distributed under a BSD license, rather than the GPL.
The question is, will they have a better experience installing software (and thus putting themselves at risk), or not installing software?
To do this, the guy blew away Windows and installed Ubuntu without telling the guy what all he did.
Pretty inexcusable.
particularly the web apps, they are pretty archaic and require IE6. But since the helpful Linux guy trashed the machine to put in his personal preference, instead of just helping the poor guy get set up, one of the guys I work with had to re-install Vista for him.
Vista? Really?
And if that was really the only problem he was having, it's possible to install IE6 on Linux. Not that I am suggesting he should've done this on his own, but then, he also couldn't have reformatted to Vista on his own.
I'd also suggest contacting the maintainers of those web apps and complaining, even if you're going to reformat. This is a guy who shouldn't have to know or care what web browser he's using. It's irresponsible of the apps in question to require a specific web browser.
I mean, to put it another way, what if he'd gotten a Mac?
you can get quite decent results by simply returning null objects (e.g. 1x1 pixel gifs, pngs, etc.) locally for certain outgoing requests
Indeed. I could even do some amount with/etc/hosts.
But on the topic of "my browser already parsed it, but I need to do $TASK on the parsed tree, why should I have to re-parse it?", here's a somewhat plan9ish fix: provide a virtual filesystem representing the HTML/XML tree, and giving other programs full read-write access.
Hmm...
At the moment, it's making more sense to me to simply rely on tools that already know how to deal with an HTML/XML DOM -- which wouldn't know WTF to do with that unpacked tree. For example, my current adblocking script for Chrome uses jQuery selectors.
Linux is no more secure from a man in the middle attack than windows, no matter what you might think.
"No matter what you might think." -- now that's a high class troll.
Let's follow your logic a bit:
The attack happens outside the OS, and the repository your Linux machine is no less vulnerable than a web browser in Windows.
Except that the web browser in Windows isn't verifying the signature of the app I'm downloading. The package manager in Linux is, for everything.
If I install a third-party repository that has its own set of keys, I'm vulnerable to a MITM exactly once, when I set it up. From then on, I am not vulnerable to a MITM.
If you think downloading a binary opens you up to MITM any more than downloading an HTML file, then you don't know enough about network security.
If you think a successful MITM on a binary is no more dangerous than a successful MITM on an HTML file, then you don't know enough about security in general.
Um... duh? The browser sandboxes the HTML. Worst they could do is display wrong or misleading information in a browser window, unless there's actually a vulnerability in the browser.
Huh? SSL stands for Secure Socket Layer, it's a secure transfer protocol which encrypts the data it sends.
SSL is also a system of certificates, implying a chain of trust.
Ubuntu repositories operate over HTTP, which stands for HyperText Transfer Protocol. Notice anything? No "secure" in there at all, because it is NOT secure.
*facepalm*
Ubuntu packages are also cryptographically signed, and checked with GnuPG. Anyone fucks with the HTTP stream, or with the CD, or with the file on disk, and apt will complain loudly and refuse to install the file.
It also has the added benefit that it's much less work for both the server and the client -- the server can be pretty much a dumb HTTP server, and the client doesn't have to deal with a crypto stream, just a signature check.
In fact, I'll bet money this is how Windows Update works.
That doesn't even make sense man.
Lots of things don't make sense when you don't bother to research them.
If you want to make sense of it, man apt-key apt-secure.
You are clearly not understanding my gripe.
Ditto.
But the repositories don't have everything, and for everything that is not in the repository installation sucks.
Nope. Try installing Chrome for Linux sometime. You pretty much point and click.
Why the hell should I have to make a Gentoo package for Ubuntu?
When did I ever suggest that?
In Windows and OSX such packages are standard.
In Ubuntu, there is a standard for them, too. It's called a deb file. Again, check out Chrome for an example of this done right.
The main difference is, on Windows and OS X, there are repositories (Microsoft Update, Apple's Software Update), but they aren't available to third-party developers.
If a Gentoo package really is the answer, why doesn't everyone who does not put their software into a repository build a Gentoo-style package?
Mostly because Portage, the Gentoo package manager, doesn't have wide adoption. Nor does Gentoo itself. So while it wouldn't be terribly difficult to do so, there's just no incentive right now.
Hell, a stupid little script that peeked inside a tarball and said "Hey, that's an application!" and popped up a little window that said "This is an application package, would you like to configure and install it?" would be brilliant, easy, and pretty much universal.
It would be close to universal. It would become universal if the developers attached a little metadata.
Get a well-supported wireless card. Most Intel cards are supported out of the box on Ubuntu -- literally, point, click, done.
or listen to music and open a video up at the same time
If it's that you want different sound sources to mix, that's one of the motivations behind PulseAudio. But I don't use it -- even the cheap sound that came in my laptop seems to mix properly in ALSA.
If it was an issue at all, I can mplayer -nosound.
or play around with some voice recognition software.
Got me there, I haven't tried that.
I want to use my computer to do stuff, too, not spend all day trying to figure out why cygwin doesn't work, or how to make my monitor keep the right aspect ratio when I change resolutions, or how to get a decent commandline running.
It's possible that my needs are different -- but then, I also don't have to fight with antivirus/antispyware/antitrojan/antiphishing/antieverything, as I know how to keep my Windows clean.
But then, part of that is taking regular disk images. Anyone know offhand how to make a Windows livecd that can be used to take a disk image of my regular Windows install, while browsing the Web and generally using my computer (everything except the Windows partition)?
Also, the annoyances you've mentioned aren't really related to installing software.
Is it really that difficult to take the configure/compile/install process and package it up into one step?
More than you'd think. But mostly, it's something to which there is no single grand unified solution. I'll give you an example of solutions which have been tried and failed miserably:.app folders (the OS X concept) originated on Unix. Unix people tend not to like them because they're bloated, and they don't make it trivial to patch one library for all applications which use it.
Making a package manager that can handle configure/make/install for any app isn't trivial, but it has been done -- see various ports systems (BSD, MacPorts), or Gentoo Portage for a better example. It's pretty trivial to build a Gentoo ebuild (package), which is ultimately a script that tells Gentoo how to download the source in a tarball (often right from SourceForge), compile it, install it, and record that installation process so it can be uninstalled later. (Oh, and it handles dependencies, too.)
Ultimately, this sucks for two big reasons: First, you still end up with breakage if you customize your system at all. And second, why should you wait hours for something like KDE to compile, when a binary distro can give you the compiled version instantly?
Everybody else can do it
By "everybody else", you obviously mean "BSD", because neither Windows nor OS X has done that. All they've done is specified a convenient binary format -- in the case of Windows, it's not even that, it's just a convention of using an exe.
Also worth mentioning: We have that too. It's called a deb. And also an rpm. And also a... hmm.
Then again, if you just release either a deb or an rpm, people from the other distro will figure out how to install it anyway. If you release it under an open license, maybe someone will even package it for you.
I suspect the main reason people don't do this is to make their app more portable -- the source is more likely to work on more platforms (including other CPU architectures) than if it was source. That's kind of the whole point of the./configure script in the first place.
Now, you may be right that it would be a good idea to have a package format that could at least compile itself on any platform. But that's a bit like PulseAudio -- trying to unify all the different package managers by adding yet another doesn't solve anything unless it absolutely mops the floor with those other package managers.
This is OSS and Linux, it's supposed to be all about choice, but the answer is to trust some corporation who, frankly, I don't know much about?
Only because you haven't researched it much. All of the Ubuntu source is online, and their process is generally transparent, as is their issue tracker, etc etc.
And if it's easy to get in to Universe, you're security argument is out the window because it won't be hard to fool whover is checking the software as it comes in.
It makes the security weaker, but not gone.
For example: Suppose I want to install Firefox. I know about Firefox already, and I know I trust it. On Windows, I would have to go to mozilla.com and download a binary -- which opens me up to a MITM attack, or to someone having compromised mozilla.com, or me mistyping the name, etc etc.
On Linux, I go to my package manager and type "firefox". I now have some assurance that, at the very least, this is the actual version of Firefox, from Mozilla, probably with some Ubuntu-specific modifications.
It's similar to SSL, only much more so.
You're making my point for me with this one. That it is more difficult does not improve security.
Except this part wasn't about security. Here, you've already downloaded a random tarball -- somewhat equivalent to downloading a random binary.
Why isn't there a script that takes care of the./configure, make, and install, and deals with any library version issues for me?
Source-based distros are essentially a collection of scripts like that. The difficulty would be making a script that works on every distro, which is why it's generally left to the distro. Essentially, what you're describing is like a Gentoo package, and if you can make a Gentoo package, why not make a package for Ubuntu?
And again, I can't remember the last time I had a library version issue, so having to./configure && make && make install isn't really more difficult to me than clicking next a bunch.
And again, this is the fallback when it isn't actually in the repository.
I want a scripted container that holds the entire package, and when run it does the configure, make, and install all at once.
You've just described a Gentoo ebuild. All it does beyond that is download the source.
The reason it is easy is because *drumroll* IT'S EASY!!! Are you honestly trying to suggest that double clicking an install file or dragging and dropping an install file is somehow only easier than the mess of configure/make/compile simply because it is familiar?
No. I'm suggesting that double-clicking an install file or dragging and dropping an install file is only easier than the convenience of using a package manager because it is familiar.
Let me give you an example of this done close to right: Rubygems. There actually aren't any install scripts, which is fine, for what this is used for. All gems get installed to predefined places that make sense. And the gem system itself is portable -- the same RubyGem can work on OS X, Linux, and most other systems, though many of them will need win32-specific versions.
Slashdot Mirror
Android is an open source OS. This means if there's a feature you need, you can add it.
I don't know if there's a central management server already built, but there's no reason you couldn't write or adapt one.
I mean, I understand why the end-user interface should be integrated, or at least well linked together.
But the whole Unix philosophy, the whole reason it works so well, is that we tend to isolate things as "do one thing and well". If I were to try to compete with Exchange, I'd use Postfix for the actual SMTP server, probably something like RoundCube for the webmail, etc. Indeed, even Postfix can be broken down into tiny, replaceable programs that communicate via pipes.
Why would I want one monolithic app that handles everything?
Anyone who thinkos otherwise hasn't spent any/enough time in a real, productive office workplace.
I have, but it was a small office.
Let me ask, then: What, specifically, is missing from Google Apps in terms of calendar/email?
Google Apps, THANK GOD, doesn't deliver. Moving from platform lock in on the desktop to another platform lock in where not only the software that I'm using but also my frakkin' USER DATA is also locked in is literally jumping out of the frying pan and into the fire.
Except that your user data is mostly available, and in open standard formats. I don't know about Docs, but the Calendars can be downloaded as iCal, and the mail can downloaded via IMAP. And in the mean time, you've got a much nicer web interface that works cross-browser, pretty much by definition cross-OS, while Exchange pretty much works with Outlook.
Can't you idiots see that Google has the makings of an evil monopoly that makes Microsoft look like your friendly local corner store.
I can, but there are several big reasons I'm not nearly as worried about Google as I am about, say, Apple (who truly does make Microsoft look like a friendly corner store, at least on mobile devices):
First, Google stuff is technically superior to Microsoft stuff in most ways that matter to me. Chrome is a better browser than IE, and before Chrome, Google sponsored Firefox.
Second, Google does seem to have a real commitment to open source and open standards. See: Summer of Code, GTalk (uses Jabber), OpenID, ODF in Google Docs, Firefox, Chrome (the browser, the OS, and 99.999% of the support needed to build it, excluding some codec support), Android, native Linux ports of most of their apps (even proprietary ones), free APIs to their proprietary services (like maps -- Microsoft wants you to call them before they'll even give you a price), Gears, O3D, Protocol Buffers...
Microsoft does some interesting open source things, but they tend to prefer their own license to more standard, proven licenses such as BSD or GPL, and they also tend to do it half-assedly -- for instance, while Google will give away a VM (as with Dalvik), Microsoft will not give the source to the .NET runtime. Since so much of what they do lately (including open source stuff) depends on .NET, this is a major roadblock. Sure, they help Mono along, but for how long?
Perhaps I'm not being fair, but Microsoft has a long history of embrace, extend, extinguish. Google, on the other hand, has a history of create/embrace, extend, set free.
Just because I can: "Random or planned" is a false dichotomy. You fail, like every watchmaker apologist before you.
To mods: Make up your minds. Either mod it funny to ridicule it, or mod it offtopic so we don't have to look at it.
Actually, web-based, free emails could be remarkably secure, if people weren't such morons about passwords.
Opera is faster and more accurate than chrome.
Citation needed.
Not that it would particularly matter to me -- Chrome is pretty accurate, and it's actually open source and easily extendable (in the developer version).
Regarding memory: First, in these tests at least, the measurements of Chrome were wrong. And second, does it matter? I have 4 gigs of RAM. It's hard to remember the last time I went into swap, and I certainly can't remember it being because of a web browser.
Regarding crashing: I don't know about you, but I'd rather have my whole browser not crash than have it be able to recover my tabs when it does. Chrome does keep a "last closed" set of tabs, but I can also comfortably run a development release -- seriously a nightly build -- as my primary browser, and I haven't seen the whole thing crash yet. I've only been able to make tabs crash with the developer tools and/or my own badly written extensions -- not that it's an excuse, just pointing out, tabs don't crash a lot.
That's more than I can say for Firefox -- Firefox (stable) crashes less often than these Chrome tabs (again, development release, lots of hacking around on my part), but I've seen the whole browser go down, taking all my tabs with it.
The developer nightlies. From my about:version:
Chromium 4.0.207.0 (Developer Build Ubuntu build 25593)
WebKit 532.0
V8 1.3.9
User Agent Mozilla/5.0 (X11; U; Linux x86_64; en-US) AppleWebKit/532.0 (KHTML, like Gecko) Chrome/4.0.207.0 Safari/532.0
I'll apply these to Chrome:
Comes with the distribution repositories and is stable, maintainable and patched
Not yet, but working on it. In the meantime, Google has its own repositories.
Has effective script control (white-listing, base-domain)
Some of this could be done with a proxy.
But seriously, I have no idea how this is going to work, if they support it at all.
Has effective ad blocking capability
I wrote an adblocker for Chrome in two days. Sure, it was for the development version of Chromium, but you get the idea -- once the extension API is released, this is just going to be a non-issue.
Also, Iron (a fork of Chromium) has an adblocker built in. I'd be surprised if they don't have some sort of script control, also.
Does surf the web and performs adequately on my system
Is there a browser that fails this?
A good tester should bookmark about 200 sites in various categories, save passwords for about 20-30 sites, have some forms saved,
How do you do that scientifically? Even eight tabs is going to raise questions of fairness -- maybe one of those pages lagged during the test?
And, you're talking about some tech journalist writing an interesting article. Where are you going to find someone willing to actually do the test you're describing? Sounds like a lot of work.
We don't care whether the browser hogs half a gig or more.
I have four gigs of RAM. What else am I using it for?
And of course, see Hurricane78 for an explanation of why that's "hogging" half a gig of RAM. Keep in mind that these are sort of theoretical maximums -- anything the browser does for you above and beyond just rendering a page is going to want its own RAM, too.
We don't care whether it can render a page correctly or makes CSS look like a 5 year old had a field day with some sharpies.
Do any of the browsers in question look like a 5 year old with sharpies?
Chrome was the overall winner, and it uses Webkit, which has historically been pretty good with web standards.
crappy servers crammed into farms that oversold their capacity hundredfold and ISPs doing the same.
So you're saying, why not slow it down even more?
I have fiber. Some sites are going to be slow, but some sites will be fast. On the fast sites, the difference between a page loading instantly, or some script executing before I can blink, is nice compared to "a few seconds".
Did he, by any chance, disable the phishing/malware protection?
And did he actually open up a packet sniffer and see those URLs go to Google? Or is it possible this was a coincidence, seeing as it's only really happened once?
What's frustrating is that in the space of all this, I also upgraded from a 32-bit Flash 9 on Ubuntu to a 64-bit Flash 10.
That Flash 9 could play fullscreen video. Not, you know, well, but it could do it. And this Flash 10 turns it into a fucking slideshow.
KDE4 has this not only built into Konqueror, but into KDE itself.
That is, I can currently do alt+f2, gg: foo, and it will open a new Konqueror window Googling for 'foo'.
It's gotten a lot better -- and for that matter, I'm posting this from the nightly. I use it as my normal browser.
That, and Google Chrome has a developer version out for Linux also.
Firebug is built-in as the Chrome Development Tools.
Adblock is available in a fork of Chrome, called Iron. It's also trivial to create an adblocking extension to Chrome itself -- and someone's done this, it's just not maintained.
Unlikely.
Firefox addons run in the same process as Firefox, likely in the same thread. Firefox tabs are similar. All it takes is one slow extension to slow down the entire experience.
Chrome, on the other hand, is implementing addons as just privileged webpages. This means that, except for the very small part of an addon that might be interacting with the current page, the addon won't block the browser -- it's mostly going to be running in a separate process. And even the content script that's running on the current page, well, there's one of those running per tab, so an extension being slow in one tab won't block another tab.
Not to mention, if you're going to implement a nice, cross-platform Firefox addon, you're doing it in Javascript/XUL. Chrome addons are Javascript/HTML. Thus, Chrome's faster Javascript engine does count here.
They also used a beta of Opera, if I recall -- and they did compare to stable versions, also.
Your ??? is "install foo", where foo is the name of whatever you wanted.
Never mind the point-and-click GUI that is add/remove programs.
most will be able to read and understand the words that appear in the folder telling them to drag the icon onto the big Applications alias icon sitting beside it.
Of the Mac users I've seen, I've only seen two ends of the spectrum.
One is highly technical. They use a Mac because they like Unix, but they also like stuff Just Working, and they like the slick, shiny interface, and all the proprietary software available. These people would be perfectly capable of installing software from MacPorts or RubyGems if they had to, and indeed, Macs are popular among the Rails community.
The other is highly non-technical. They like a Mac because it's what they've always used, and they have no idea what goes on under the hood. It's these people who tend to ignore the instructions and focus on the pictures.
And it's here that Linux is actually easier -- because if someone includes a link to an apt-url, or to a deb, they can point and click, and it'll work, and there is no way they can screw it up and have it be half-installed as a disk image somewhere.
Follow the instructions... like the one that says "drag to your applications folder" and has the alias right there.
You're expecting the average user -- the one we're talking about in this discussion -- to read, instead of "Oh, there's my app! *click*"
Trust me, I've seen that particular instruction in a Firefox DMG that was being used in just this manner.
if your installer is in a Zip file, you open it and run it.
Unless the zipfile is of the program itself, in which case, you make a folder, drag the contents of the zipfile to that folder, and run it. And technically, you probably should be putting that folder somewhere in Program Files, and making it owned by whatever usually owns Program Files, and, you know, all the stuff that an installer normally does for you.
it's still corporate. As is Fedora and Red Hat and Debian.
You may well have a point with Ubuntu, but which corporation runs Debian?
When I mention "driver" to someone and they give me a funny look and say "What's a driver?", do you really think they will have an easy time of using Linux?
I suspect they already do, with Google and tons of websites they use, and possibly with their phone. As for the desktop, absolutely, yes.
Perhaps they will have a difficult time installing Linux, but they would very likely have an easy time of using it.
Put another way: Is that the kind of person who generally installs software on their computer? Is it the kind of person you want installing software on their computer?
you'll spend 20 minutes just trying to get them to open a terminal so you can give them the commands to fix whatever issues they have that cannot be fixed via the GUI.
After which I'll spend ten seconds typing those commands into an IM window, and having them copying and pasting them into that terminal.
And that assumes you're right about those 20 minutes. Most people I know can handle "Click the word Applications in the upper left of the screen. Then click Accessories. Then click Terminal."
But again, there's still that large assumption that such an issue will come up at all. In my experience, issues come up for people on Windows that require such calls much more frequently than on Linux.
What you think someone should or should not be allowed to do is irrelavent and imposing that on others frankly goes against the whole Linux and free software philosophy.
I beg to differ. If that was really the case, Linux would likely be distributed under a BSD license, rather than the GPL.
The question is, will they have a better experience installing software (and thus putting themselves at risk), or not installing software?
To do this, the guy blew away Windows and installed Ubuntu without telling the guy what all he did.
Pretty inexcusable.
particularly the web apps, they are pretty archaic and require IE6. But since the helpful Linux guy trashed the machine to put in his personal preference, instead of just helping the poor guy get set up, one of the guys I work with had to re-install Vista for him.
Vista? Really?
And if that was really the only problem he was having, it's possible to install IE6 on Linux. Not that I am suggesting he should've done this on his own, but then, he also couldn't have reformatted to Vista on his own.
I'd also suggest contacting the maintainers of those web apps and complaining, even if you're going to reformat. This is a guy who shouldn't have to know or care what web browser he's using. It's irresponsible of the apps in question to require a specific web browser.
I mean, to put it another way, what if he'd gotten a Mac?
you can get quite decent results by simply returning null objects (e.g. 1x1 pixel gifs, pngs, etc.) locally for certain outgoing requests
Indeed. I could even do some amount with /etc/hosts.
But on the topic of "my browser already parsed it, but I need to do $TASK on the parsed tree, why should I have to re-parse it?", here's a somewhat plan9ish fix: provide a virtual filesystem representing the HTML/XML tree, and giving other programs full read-write access.
Hmm...
At the moment, it's making more sense to me to simply rely on tools that already know how to deal with an HTML/XML DOM -- which wouldn't know WTF to do with that unpacked tree. For example, my current adblocking script for Chrome uses jQuery selectors.
Linux is no more secure from a man in the middle attack than windows, no matter what you might think.
"No matter what you might think." -- now that's a high class troll.
Let's follow your logic a bit:
The attack happens outside the OS, and the repository your Linux machine is no less vulnerable than a web browser in Windows.
Except that the web browser in Windows isn't verifying the signature of the app I'm downloading. The package manager in Linux is, for everything.
If I install a third-party repository that has its own set of keys, I'm vulnerable to a MITM exactly once, when I set it up. From then on, I am not vulnerable to a MITM.
If you think downloading a binary opens you up to MITM any more than downloading an HTML file, then you don't know enough about network security.
If you think a successful MITM on a binary is no more dangerous than a successful MITM on an HTML file, then you don't know enough about security in general.
Um... duh? The browser sandboxes the HTML. Worst they could do is display wrong or misleading information in a browser window, unless there's actually a vulnerability in the browser.
Huh? SSL stands for Secure Socket Layer, it's a secure transfer protocol which encrypts the data it sends.
SSL is also a system of certificates, implying a chain of trust.
Ubuntu repositories operate over HTTP, which stands for HyperText Transfer Protocol. Notice anything? No "secure" in there at all, because it is NOT secure.
*facepalm*
Ubuntu packages are also cryptographically signed, and checked with GnuPG. Anyone fucks with the HTTP stream, or with the CD, or with the file on disk, and apt will complain loudly and refuse to install the file.
It also has the added benefit that it's much less work for both the server and the client -- the server can be pretty much a dumb HTTP server, and the client doesn't have to deal with a crypto stream, just a signature check.
In fact, I'll bet money this is how Windows Update works.
That doesn't even make sense man.
Lots of things don't make sense when you don't bother to research them.
If you want to make sense of it, man apt-key apt-secure.
You are clearly not understanding my gripe.
Ditto.
But the repositories don't have everything, and for everything that is not in the repository installation sucks.
Nope. Try installing Chrome for Linux sometime. You pretty much point and click.
Why the hell should I have to make a Gentoo package for Ubuntu?
When did I ever suggest that?
In Windows and OSX such packages are standard.
In Ubuntu, there is a standard for them, too. It's called a deb file. Again, check out Chrome for an example of this done right.
The main difference is, on Windows and OS X, there are repositories (Microsoft Update, Apple's Software Update), but they aren't available to third-party developers.
If a Gentoo package really is the answer, why doesn't everyone who does not put their software into a repository build a Gentoo-style package?
Mostly because Portage, the Gentoo package manager, doesn't have wide adoption. Nor does Gentoo itself. So while it wouldn't be terribly difficult to do so, there's just no incentive right now.
Hell, a stupid little script that peeked inside a tarball and said "Hey, that's an application!" and popped up a little window that said "This is an application package, would you like to configure and install it?" would be brilliant, easy, and pretty much universal.
It would be close to universal. It would become universal if the developers attached a little metadata.
But it wouldn't be universal. Most packages ar
all I'm trying to do is use my wireless internet
Get a well-supported wireless card. Most Intel cards are supported out of the box on Ubuntu -- literally, point, click, done.
or listen to music and open a video up at the same time
If it's that you want different sound sources to mix, that's one of the motivations behind PulseAudio. But I don't use it -- even the cheap sound that came in my laptop seems to mix properly in ALSA.
If it was an issue at all, I can mplayer -nosound.
or play around with some voice recognition software.
Got me there, I haven't tried that.
I want to use my computer to do stuff, too, not spend all day trying to figure out why cygwin doesn't work, or how to make my monitor keep the right aspect ratio when I change resolutions, or how to get a decent commandline running.
It's possible that my needs are different -- but then, I also don't have to fight with antivirus/antispyware/antitrojan/antiphishing/antieverything, as I know how to keep my Windows clean.
But then, part of that is taking regular disk images. Anyone know offhand how to make a Windows livecd that can be used to take a disk image of my regular Windows install, while browsing the Web and generally using my computer (everything except the Windows partition)?
Also, the annoyances you've mentioned aren't really related to installing software.
Is it really that difficult to take the configure/compile/install process and package it up into one step?
More than you'd think. But mostly, it's something to which there is no single grand unified solution. I'll give you an example of solutions which have been tried and failed miserably: .app folders (the OS X concept) originated on Unix. Unix people tend not to like them because they're bloated, and they don't make it trivial to patch one library for all applications which use it.
Making a package manager that can handle configure/make/install for any app isn't trivial, but it has been done -- see various ports systems (BSD, MacPorts), or Gentoo Portage for a better example. It's pretty trivial to build a Gentoo ebuild (package), which is ultimately a script that tells Gentoo how to download the source in a tarball (often right from SourceForge), compile it, install it, and record that installation process so it can be uninstalled later. (Oh, and it handles dependencies, too.)
Ultimately, this sucks for two big reasons: First, you still end up with breakage if you customize your system at all. And second, why should you wait hours for something like KDE to compile, when a binary distro can give you the compiled version instantly?
Everybody else can do it
By "everybody else", you obviously mean "BSD", because neither Windows nor OS X has done that. All they've done is specified a convenient binary format -- in the case of Windows, it's not even that, it's just a convention of using an exe.
Also worth mentioning: We have that too. It's called a deb. And also an rpm. And also a... hmm.
Then again, if you just release either a deb or an rpm, people from the other distro will figure out how to install it anyway. If you release it under an open license, maybe someone will even package it for you.
I suspect the main reason people don't do this is to make their app more portable -- the source is more likely to work on more platforms (including other CPU architectures) than if it was source. That's kind of the whole point of the ./configure script in the first place.
Now, you may be right that it would be a good idea to have a package format that could at least compile itself on any platform. But that's a bit like PulseAudio -- trying to unify all the different package managers by adding yet another doesn't solve anything unless it absolutely mops the floor with those other package managers.
And agai
Unfortunately, that's mostly a Flash bug, and not a Linux bug. For proof, take the same video and play it in mplayer or vlc -- no pauses.
This is OSS and Linux, it's supposed to be all about choice, but the answer is to trust some corporation who, frankly, I don't know much about?
Only because you haven't researched it much. All of the Ubuntu source is online, and their process is generally transparent, as is their issue tracker, etc etc.
And if it's easy to get in to Universe, you're security argument is out the window because it won't be hard to fool whover is checking the software as it comes in.
It makes the security weaker, but not gone.
For example: Suppose I want to install Firefox. I know about Firefox already, and I know I trust it. On Windows, I would have to go to mozilla.com and download a binary -- which opens me up to a MITM attack, or to someone having compromised mozilla.com, or me mistyping the name, etc etc.
On Linux, I go to my package manager and type "firefox". I now have some assurance that, at the very least, this is the actual version of Firefox, from Mozilla, probably with some Ubuntu-specific modifications.
It's similar to SSL, only much more so.
You're making my point for me with this one. That it is more difficult does not improve security.
Except this part wasn't about security. Here, you've already downloaded a random tarball -- somewhat equivalent to downloading a random binary.
Why isn't there a script that takes care of the ./configure, make, and install, and deals with any library version issues for me?
Source-based distros are essentially a collection of scripts like that. The difficulty would be making a script that works on every distro, which is why it's generally left to the distro. Essentially, what you're describing is like a Gentoo package, and if you can make a Gentoo package, why not make a package for Ubuntu?
And again, I can't remember the last time I had a library version issue, so having to ./configure && make && make install isn't really more difficult to me than clicking next a bunch.
And again, this is the fallback when it isn't actually in the repository.
I want a scripted container that holds the entire package, and when run it does the configure, make, and install all at once.
You've just described a Gentoo ebuild. All it does beyond that is download the source.
The reason it is easy is because *drumroll* IT'S EASY!!! Are you honestly trying to suggest that double clicking an install file or dragging and dropping an install file is somehow only easier than the mess of configure/make/compile simply because it is familiar?
No. I'm suggesting that double-clicking an install file or dragging and dropping an install file is only easier than the convenience of using a package manager because it is familiar.
Let me give you an example of this done close to right: Rubygems. There actually aren't any install scripts, which is fine, for what this is used for. All gems get installed to predefined places that make sense. And the gem system itself is portable -- the same RubyGem can work on OS X, Linux, and most other systems, though many of them will need win32-specific versions.