Safari 5.0.5

[+Frank B. Subscriber²]

This update is recommended for all Safari users and includes the latest security updates.

For information on the security content of this update, please visit this website: http://support.apple.com/kb/HT1222.

Download: via Apple Software Update or from the Safari website.

Restart required.

[ObiWanToby]

Good ol Windows and OS X always requiring reboots for browsers.

[.Neo]

Good ol Windows and OS X always requiring reboots for browsers.

Because Safari relies on more than just that .app package sitting in /Applications... For some reason people have a hard time understanding that.

[Quillz]

Good ol Windows and OS X always requiring reboots for browsers.

Safari updates also update the internal WebKit version, which affects the system kernel and requires a reboot. This is why IE updates on Windows need reboots, too, as well as most Linux distros when they install browser and/or kernel updates.

[OuchOfDeath]

How the hell does webkit affect the kernel? That makes no sense at all. Also as of IE9 reboots are no longer required because Microsoft has finally started coding their sh*t properly.

Linux has never needed a restart for a browser. Ever. It doesn't make any sense, and therefore has never happened. In fact you can upgrade the browser (or literally any program at all) while it's still running, and all you need to do is turn it off and run it again and the newer version runs seamlessly.

[+Frank B. Subscriber²]

How the hell does webkit affect the kernel? That makes no sense at all. Also as of IE9 reboots are no longer required because Microsoft has finally started coding their sh*t properly.

Linux has never needed a restart for a browser. Ever. It doesn't make any sense, and therefore has never happened. In fact you can upgrade the browser (or literally any program at all) while it's still running, and all you need to do is turn it off and run it again and the newer version runs seamlessly.

WebKit does not affect the OS X kernel at all. Its library is used by various OS components however, hence the need for a reboot.

Also, IE 9 x86 still forces you to reboot after installing it.

[.Neo]

Linux has never needed a restart for a browser. Ever. It doesn't make any sense, and therefore has never happened. In fact you can upgrade the browser (or literally any program at all) while it's still running, and all you need to do is turn it off and run it again and the newer version runs seamlessly.

Most Linux distros, such as Ubuntu, rely on third-party browsers that aren't an integral part of the OS. Updating Firefox and Chrome on Mac OS X don't require a restart for that very reason either: They are stand-alone entities. Safari updates aren't just Safari updates, they also update the WebKit framework used by many of Mac OS X' build-in applications, services and sometimes even third-party applications. Hence why the operating system needs to restart.

Crucial system updates on Linux requite a restart all the same. On Mac OS X WebKit is such a crucial system update as it extends well beyond Safari alone. As such it makes a lot of sense.

Hell, you don't have to take our word for it. You can see what a "simple" Safari update breaks down into yourself:

[OuchOfDeath]

Most Linux distros, such as Ubuntu, rely on third-party browsers that aren't an integral part of the OS. Updating Firefox and Chrome on Mac OS X don't require a restart for that very reason either: They are stand-alone entities. Safari updates aren't just Safari updates, they also update the WebKit framework used by many of Mac OS X' build-in applications, services and sometimes even third-party applications. Hence why the operating system needs to restart.

Crucial system updates on Linux requite a restart all the same. On Mac OS X WebKit is such a crucial system update as it extends well beyond Safari alone. As such it makes a lot of sense.

Hell, you don't have to take our word for it. You can see what a "simple" Safari update breaks down into yourself:

Lets not attribute the wrong concepts to the wrong problems here. There's a lot to break down. I'll start with this:

Updating a user-land framework, especially one as simple as webkit, should not require a restart. Period. There is -zero- kernel-level code involved. If there is then that is unbelievably screwed up and an even bigger problem. There are no ifs ands or buts. If the operating system makes you restart over userland-level code it's redundant and a waste of time. That or the operating system has not been designed correctly, and in most cases that has been true for Windows and for OS X as well it seems. In Windows (since Vista I believe) and Linux you can INSTALL THE VIDEO CARD DRIVER WITHOUT A RESTART. Please try to picture this. This is the most advanced driver in the operating system (which is kernel-level code, remember), and it can be installed without a restart. Why in the world does a simple library/framework (yes, it is unbelievably simple userland code compared to a kernel-level video driver) require a restart? Why? Do you not see the giant contradiction and complete lack of logic here?

Furthermore, I'd like to correct you on the fact that Linux requires a restart all the time. This is incorrect. Linux only needs a restart under literally a single condition. That is when you replace the kernel. You can install as many drivers as you like without a restart. You may have run Ubuntu and Fedora before and you are right in that they they do indeed ask for a restart when you install a graphics driver, however these restarts are not required at all for the driver to work, and frankly I don't know why the distros opt for a restart. I have a feeling it has something to do with the ATI driver as it's historically been a mess, but that's something else entirely.

On a slightly unrelated topic, the whole GUI framework in Linux (aka your session/login) does need a restart(of itself, not the computer) in its current form when a graphics driver is installed, and this is likely attributed to the fact that it is extremely old and not built from the ground up to be this modular, aka it's not coded properly. Windows however does not even require a session restart. You just get a few seconds of your monitor blinking and there you go. You have your driver. Windows is doing things quite a bit better here which is great to see. But back to the topic at hand.

I think it falls on you to explain to everyone here why userland code such as webkit requires a restart when it logically does not make sense. In this day and age we have kernel-level code of the highest complexity such as gpu drivers installed and updated on the fly, and you're somehow arguing that a browser engine of all things requires a restart? I'll emphasize again. Webkit = user-level code. Drivers = kernel-level code. You see how upside down this whole thing is?

[.Neo]

I think it falls on you to explain to everyone here why userland code such as webkit requires a restart when it logically does not make sense. In this day and age we have kernel-level code of the highest complexity such as gpu drivers installed and updated on the fly, and you're somehow arguing that a browser engine of all things requires a restart? I'll emphasize again. Webkit = user-level code. Drivers = kernel-level code. You see how upside down this whole thing is?

Excuse me, it falls on me to explain this? :laugh: Mac OS X isn't my responsibility so in the end I don't have to explain a thing. What Apple's exact reasons are to force a restart is anyone's guess. Could be it's nothing strictly technical but they try to create an environment where they make the risk of an update going wrong as small as possible. While I was impressed at first, I've seen those video card driver installs on Windows fail quite a few times forcing you to restart anyway. And when it comes to Linux I find myself restarting for updates more than enough as well. Maybe not for a browser and its system frameworks (doesn't apply to Firefox and Chrome to begin with but okay) but for other stuff. In the end I don't care what OS requires me to restart for what, it's a restart all the same.

Furthermore, I'd like to correct you on the fact that Linux requires a restart all the time. This is incorrect.

Really? Last time I had to restart Ubuntu for updates was a week ago. For the sake of this discussion I started up Ubuntu just now, ran the Update Manager and find myself having to restart again. Honestly, I don't mind but arguing here restarting Linux for software updates only happens once in a blue moon is nonsense. I don't care what parts of Ubuntu require restarts and which don't and if it's strictly needed from a technical standpoint or not, again, a restart is a restart and it happens quite frequently on that side as well.

It amazes me you get so worked up over a restart that takes, like what, 30 seconds? When it comes to both Mac OS X and Ubuntu I get the choice to hold off the restart until it suits me. So I really couldn't care less about this. The only thing that does ****es me off beyond believe is/was Windows' way of harassing me over and over again with an annoying pop-up balloon asking me to restart if I didn't choose to do so right away. Hopefully Microsoft finally got rid of that behavior once and for all because it's annoying as hell.

[OuchOfDeath]

Yes, if you make a statement like the one you made it falls on you to back it up with some sort of reasoning. You were defending the need for a restart over a simple userland library, and I noticed it didn't make logical sense, and provided you with the very good example of graphics drivers being loaded and updated on the fly.

Apple doesn't have any reasons. There are no good reasons for a restart(single exception, kernel replacement), let alone a restart over userland code. Microsoft have a huge limitation somewhere in their design that causes this problem to be more prominent and harder to code for. It's the "this file is in use by another program" design where files become locked to a process. Linux does not have this limitation, but I don't know about OS X. Regardless of this limitation Microsoft is doing an excellent job of modularizing their operating system and coding it properly to remove all the restarts that should not have been there in the first place. Being able to update your graphics driver without even having to log out is excellent design, and it should be implemented across the board.

Really? Last time I had to restart Ubuntu for updates was a week ago. For the sake of this discussion I started up Ubuntu just now, ran the Update Manager and find myself having to restart again. Honestly, I don't mind but arguing here restarting Linux for software updates only happens once in a blue moon is nonsense. I don't care what parts of Ubuntu require restarts and which don't, again, a restart is a restart and it happens quite frequently on that side as well.

I... already addressed that. Yes Ubuntu prompts you for a restart after installing a graphics driver. No it's not required. I've installed and updated my graphics driver in Linux a thousand times without restarting(NVIDIA, ATI's driver is ugly and may cause problems). See, what you do is you ignore the "YOU HAVE TO RESTART YOUR COMPUTER" dialog because it's nonsense, and you restart the xserver (the entire GUI), and then the driver is loaded and working. The restart is unnecesary, and i already said it may be because of ATI's driver being a mess that those dialogs are there. The single time Linux actually needs to restart is after upgrading the kernel.

EDIT: I'm not getting "worked up", just so you know. I enjoy explaining operating system design and being challenged when I'm wrong. It's a forum. It's discussion.

[.Neo]

Yes, if you make a statement like the one you made it falls on you to back it up with some sort of reasoning. You were defending the need for a restart over a simple userland library, and I noticed it didn't make logical sense, and provided you with the very good example of graphics drivers being loaded and updated on the fly.

My reasoning still stands. Safari updates exist out of more than the application alone with many applications and services depending on it. Forcing a restart is probably the quickest and safest route to guarantee the install goes right and that everything is making use of it afterwards.

Apple doesn't have any reasons.

That's nonsense.

Regardless of this limitation Microsoft is doing an excellent job of modularizing their operating system and coding it properly to remove all the restarts that should not have been there in the first place. Being able to update your graphics driver without even having to log out is excellent design, and it should be implemented across the board.

Like I said, I've seen this particular thing go south many times which makes me wonder if it's actually the right path to take. Banning out restarts at the risk of an update not being executed right isn't a tradeoff I'm willing to make. Even Internet Explorer 9's installer can fail at shutting down all involved services prior to replacing them with never versions requiring you to do a restart all the same.

I... already addressed that. Yes Ubuntu prompts you for a restart after installing a graphics driver. No it's not required. I've installed and updated my graphics driver in Linux a thousand times without restarting(NVIDIA, ATI's driver is ugly and may cause problems). See, what you do is you ignore the "YOU HAVE TO RESTART YOUR COMPUTER" dialog because it's nonsense, and you restart the xserver (the entire GUI), and then the driver is loaded and working. The restart is unnecesary, and i already said it may be because of ATI's driver being a mess that those dialogs are there. The single time Linux actually needs to restart is after upgrading the kernel.

Again, when I see a "You have to restart" dialog I follow its advice. If not directly, then at a later time. The excuse you're giving for Ubuntu can be applied to Mac OS X all the same: An installer will prompt me to restart when it's not actually necessary, the software is already there ready to be used. These things happen on both sides.

[OuchOfDeath]

My reasoning still stands. Safari updates exist out of more than the application alone with many applications and services depending on it. Forcing a restart is probably the quickest and safest route to guarantee the install goes right and that everything is making use of it afterwards.

Sure, it's the quickest and safest route if you coded your operating system the wrong way and ended up in a situation like this where a re-design to make it more modular is a ton of work and new code. However when you compare both methods side by side neither is "safer" than the other. The modular on-the-fly design is of course faster since it avoids having to restart. Now this last comment I don't have any concrete knowledge to back it up with, but I believe modular designs are more intuitive to the coders themselves since they are, well... modular. There's a clear design in-place. However that's another topic.

The bottom line is neither design is safer than the other. Apple is simply being lazy. That is the only logical conclusion one can come to. It's bad design that they have not gotten around to fixing yet.

That's nonsense.

It's logic. You have no room to argue on this. The only reason to have a non-modular design like the one here is because it's easier to code for. That's it. The only reason it's like that is because it's already there and Apple hasn't bothered to change it. It's not the correct way to go about things.

Like I said, I've seen this particular thing go south many times which makes me wonder if it's actually the right path to take. Banning out restarts at the risk of an update not being executed right isn't a tradeoff I'm willing to make. Even Internet Explorer 9's installer can fail at shutting down all involved services prior to replacing them with never versions requiring you to do a restart all the same.

This is the mistake you're making. You are going under the assumption that somehow the "restart the computer" design is safer. It's not. The only reason you think it's safer is because you've been exposed to ****ty coding practices, and when a company tries to code something properly the first time they screw it up. This is bad logic. There have been countless times in history where replacing old systems with new ones initially went horribly wrong, simply because implementing a new system requires you to... implement it from scratch. It has nothing to do with the new system's actual design being an inferior one. It's the mistake in logic you're making here.

Again, when I see a "You have to restart" dialog I follow its advice. If not directly, then at a later time. The excuse you're giving for Ubuntu can be applied to Mac OS X all the same: An installer will prompt me to restart when it's not actually necessary and the software is already there ready to be used. These things happen on both sides.

No. Now you're just being... dumb, sorry to say. I'm not making an excuse. I'm making a logical argument that you're not even grasping. Linux can update graphics driver without restart. OS X can't. I don't care about any dialogs you've seen, I'm talking about the actual DESIGN. Lets say it like this: "Linux can physically update the graphics driver without restarting. OS X can't."

I don't know what the hell you're arguing anyway. Windows can do it all without even logging out. It's the proper way of going about things. Modular design is better. Simple as that.

[.Neo]

It's logic. You have no room to argue on this. The only reason to have a non-modular design like the one here is because it's easier to code for. That's it. The only reason it's like that is because it's already there and Apple hasn't bothered to change it. It's not the correct way to go about things.

Fact is you can't possibly know what Apple's reasoning behind this is. Doesn't matter how you twist and turn things around. Where you're getting it from that Mac OS X doesn't have a modular design I have no idea. As far as I know it does.

This is the mistake you're making. You are going under the assumption that somehow the "restart the computer" design is safer. It's not. The only reason you think it's safer is because you've been exposed to ****ty coding practices, and when a company tries to code something properly the first time they screw it up. This is bad logic. There have been countless times in history where replacing old systems with new ones initially went horribly wrong, simply because implementing a new system requires you to... implement it from scratch. It has nothing to do with the new system's actual design being an inferior one. It's the mistake in logic you're making here.

What I'm talking about has nothing to do with the technical side of things. It's about protecting the system from the end-user screwing up by disabling access to the rest of the OS while the update is underway. This is exactly what you see happening on Windows: For some reason the installer can't quit all crucial services (Internet Explorer 9 installer for example) or the end-user launches an application so Windows has to restart anyway to complete the install process.

No. Now you're just being... dumb, sorry to say. I'm not making an excuse. I'm making a logical argument that you're not even grasping. Linux can update graphics driver without restart. OS X can't. I don't care about any dialogs you've seen, I'm talking about the actual DESIGN. Lets say it like this: "Linux can physically update the graphics driver without restarting. OS X can't."

I don't know what the hell you're arguing anyway. Windows can do it all without even logging out. It's the proper way of going about things. Modular design is better. Simple as that.

What is it with people on forums that they can't have a normal discussion without turning it into a ****ing contest? Anyway, at this point Mac OS X is already capable of installing drivers without the need for a restart. Printer/scanner drivers being an example.

Whether Mac OS X is fully incapable of installing graphics drivers the same way I'm not sure. I never had to install them manually. Beyond that I couldn't care less about DESIGN if in the end Ubuntu still prompts me to restart. The capabilities of an operating system are only as good as the software that makes use of them. Apparently that cooperation isn't working too well on Ubuntu at this point which results in the same thing for the end-user: A restart.

[strelok1911]

Updated just fine on my Hackintosh.

[torrentthief]

what score does this get on http://html5test.com/index.html

[OuchOfDeath]

Printer drivers are something quite different. Their complexity level is not anywhere near that of most other device drivers. As for the rest of your arguments, I just have nothing more to say. I'm making clear logical arguments that you counter with nonsensical statements. This is going nowhere. You simply don't know what the hell you're talking about. You don't even know what rings are do you? Here, I'll copy a statement from wikipedia:

"In computer science, hierarchical protection domains,[1][2] often called protection rings, are a mechanism to protect data and functionality from faults (fault tolerance) and malicious behaviour (computer security). This approach is diametrically opposite to that of capability-based security."

See, there is an entire science behind all of this, and science is final. You do not argue with science. Science is truth, and one just looks like idiot arguing against something they don't understand. When I say Apple is doing things wrong my statement is backed by logical and scientific knowledge, such as the one I provided. There is no room to argue, and if you don't understand these concepts, which I'm almost certain you don't, I'd just stop commenting and start reading instead.

[.Neo]

Sure let's just agree on that your point of view is the only one. (Y)

[OuchOfDeath]

There are no points of view when it comes to scientific fact. Fact is fact. Logic is logic. You're simply not aware of the underlying systems of any of this. Anyway agreed. No point in continuing.

[.Neo]

There are no points of view when it comes to scientific fact. Fact is fact. Logic is logic.

Except so far you haven't proofed anything yet when it comes down to Mac OS X itself.

[OuchOfDeath]

Except so far you haven't proofed anything yet when it comes down to Mac OS X itself.

Of course I haven't. I've had to depend on you for any OS X information, and you don't have any, so a part of the discussion has been assumptions. However many of the concepts discussed are OS-independent, such as the really simple usermode and kernel mode concepts (and the complete contradiction in your arguments saying webkit needs a restart when we have graphics drivers that don't, therefore by logical deduction webkit doesn't need a restart either) and you're not grasping those either.

[.Neo]

Of course I haven't. I've had to depend on you for any OS X information, and you don't have any, so a part of the discussion has been assumptions. However many of the concepts discussed are OS-independent, such as the really simple usermode and kernel mode concepts (and the complete contradiction in your arguments saying webkit needs a restart when we have graphics drivers that don't, therefore by logical deduction webkit doesn't need a restart either) and you're not grasping those either.

Never did I say the OS strictly has to restart in order to complete the WebKit installation. What I said was it could be the quickest method given how many services and applications reply on it. You're completely unwilling to differentiate between the purely technical capabilities of an OS and what's actually practical.

[OuchOfDeath]

Never did I say the OS strictly has to start in order to complete the WebKit installation. What I said was it could be the quickest method given how many services and applications reply on it. You're completely unwilling to differentiate between the purely technical capabilities of an OS and what's actually practical.

I'm absolutely willing to consider what's practical, and that's what I am doing. Let's do a bit of math here. What is faster? Have the computer reboot and then have it start up all those services? Or restart all those services (which happens at the same speed or faster than starting them up in the first place)? See how removing one entire step that is actually by far the step that takes the most time is universally(both technically and practically) the better way to go about things? There is no fancy complexity anywhere that would make a restart actually faster. That's never true.

You're absolutely unwilling to... well, I can't make it any clearer than that. If you can't understand that one step is faster than two steps then I don't know what could make you start thinking properly.

[.Neo]

I'm absolutely willing to consider what's practical, and that's what I am doing. Let's do a bit of math here. What is faster? Have the computer reboot and then have it start up all those services? Or restart all those services (which happens at the same speed or faster than starting them up in the first place)? See how removing one entire step that is actually by far the step that takes the most time is universally(both technically and practically) the better way to go about things? There is no fancy complexity anywhere that would make a restart actually faster. That's never true.

You're absolutely unwilling to... well, I can't make it any clearer than that. If you can't understand that one step is faster than two steps then I don't know what could make you start thinking properly.

The Internet Explorer 9 installer does exactly what you want it to do: It tries to shut down all services that rely on it in order to prevent a restart, tries to Install Explorer 9 and all underlying frameworks and then tries to finish the installation. It's perfect on paper. However, this is what actually happens in practice for many: For whatever reason something prevents the installer from shutting everything down, as such is unable to install everything properly and has to do a restart regardless in order to finish. So now you had to wait for the installer to quit all services, wait for a failed install attempt to finish, restart Windows and go through the installer process again. In the meantime I had to go through several prompts that something didn't went right.

Microsoft's method is undoubtedly faster if and when everything goes right. I'm not arguing that with you. Unfortunately that isn't always the case. Apparently Apple chooses not to take that risk and shuts everything down before attempting to install to begin with, resulting in a faster and smoother installation process.

[OuchOfDeath]

The Internet Explorer 9 installer does exactly what you want it to do: It tries to shut down all services that rely on it in order to prevent a restart, tries to Install Explorer 9 and all underlying frameworks and then tries to finish the installation. It's perfect on paper. However, this is what actually happens in practice for many: For whatever reason something prevents the installer from shutting everything down, as such is unable to install everything properly and has to do a restart regardless in order to finish.

Microsoft's method is undoubtedly faster if and when everything goes right. Unfortunately that isn't always the case. Apparently Apple chooses not to take that risk and shuts everything down before attempting to install to begin with.

Microsoft's method is faster, and it's the correct way to go about things. It's not implemented properly as issues as you mentioned do arise. There is no "risk" one has to take if it's implemented properly. It has the exact same risk ratio as a properly implemented restart-update sequence. This is backed by logical fact of the technical sides of the designs themselves. Apple simply did not implement the correct system here. It's simply that. They could have one that is modular and done on the fly and runs flawlessly without issue. This can be achieved just as a restart update sequence is achieved all the time. Apple just hasn't done it. That's all there is to it.

[.Neo]

Microsoft's method is faster, and it's the correct way to go about things. It's not implemented properly as issues as you mentioned do arise. There is no "risk" one has to take if it's implemented properly.

That's a pretty big "IF". So far we've seen that it just doesn't work 100% yet. Until that time Apple's old and proven method remains the smoothes experience as far as I'm concerned. You keep going on about a picture perfect situation where everything works exactly the way as intended. That's just not how things work in real life. This forum alone shows that there's an insanely high rate of Internet Explorer 9 installations fail at preventing a restart, which in the end takes up more time than Apple's solution.

You're completely unwilling to take into account Apple is more than capable of coming up with a similar installation process for Safari on Mac OS X, but that they haven't done so for the exact above reason: There simply too many variables that can cause things to go south.