Tufte's work is a great foundation for people who don't know anything about presenting data effectively. I know several people who have been to his seminars, one of whom is a CIO who has since made Tufte's books required reading for most of the people who report to him.
Like the other poster here said, Tufte's stuff isn't that useful for really complex multidimensional data. His stuff is more oriented toward presenting relatively simple stuff in ways that are readable and aren't misleading. On those two subjects, his stuff is very good, and the basic principles all still apply to the more complex stuff the submitter's asking about. The basic principles Tufte presents just don't suffice for high dimensional visualization.
Web sites that require IE are becoming rarer, but they're still around, and usually in the most annoying places. For example, my health insurance provider (which was chosen for me by my employer) has an IE only website. My parents' bank's website requires IE. Several websites I've been required to use during classes at a public university only work fully with IE. If Microsoft had never illegally established their now-eroding web browser monopoly, these websites would have never been developed to require a specific web browser. Firefox hit 1.0 more than 4 years ago, but I still have to keep a Windows virtual machine lying around on my hard drive so that I can use IE. The web definitely hasn't recovered from the harm dealt to it by Microsoft's abuse of monopoly.
The point of all of this is to prevent Microsoft from killing off third party products and companies by bundling free alternatives to the third party products with Windows, thus fixing the price at $0. Microsoft shouldn't have the power to fix prices like that, especially when their abuse of that power bankrupts companies that provided a competitive environment that encourages innovation.
Netscape is the perfect example of why Microsoft's monopoly needs to be policed. After Microsoft killed Netscape, the web browser market stagnated for years, and Microsoft had de facto control over web standards, which resulted in their being no coherent or well-defined standard for building web pages, which enabled Microsoft to create almost as much vendor lock-in as they have with document file formats.
The HTML rendering components should be made into a redistributable package, like the.NET framework. That way, anybody distributing software that needs HTML rendering capabilities can include the installer for the MS HTML engine. This arrangement would also encourage application developers to more carefully consider why and how they add HTML rendering to their app, and whether they should use some other platform, such as XULRunner or Webkit.
That's because ftp.exe was ported almost directly over from BSD, along with the rest of the windows TCP/IP stack. (The internals of the TCP/IP stack have obviously been changed a lot from what they got from BSD, but the utilities like ftp and arp haven't changed much, though Microsoft has buried the Berkeley copyright statements deeper over the years.)
Doesn't WMP include codecs that are covered by patents for which there are licensing fees? If so, the consumer should certainly be able to save money by opting out of licensing the proprietary formats.
There's been plenty of complaining about Microsoft's bundling of a software firewall, though not nearly as much as Microsoft selling security software in a fairly blatant protection racket. The disk defragmenting software was also controversial at one time. The rest of the apps you listed, with the exception of Explorer, can all be completely removed or ignored by the user. (For example, you can easily set it so that all text files open with OO.o instead of Notepad, and then you'll never see Notepad again. You can't do this for all html files and links.)
No, they're arguing that OEMs should be able to pick what browser they want installed. And no, you don't "need" a broswer at install time - most ISPs supply CDs and many of those contain browsers.
... and all of those browsers are very thin wrappers around IE, and can't be modified to use eg. Gecko or WebKit as a back-end.
For me, the big problem with the IE bundling is that many apps (such as email clients) will open html links in IE even if Firefox is the "default" browser. If Microsoft were serious about providing a way to set a default browser, they'd redirect those links to Firefox.
For comparison with another proprietary OS, specifically, Mac OS X: many first-party and built-in apps (help viewer, XCode's doc browser, and Mail.app) use Webkit to display html, but I've never run across an app that, when opening a link in an external browser, insists upon Safari. This may simply be an accident due to the fact that Safari came along long after OS X, but Apple's OS is clearly more accommodating of the user's wishes in this regard.
As I recall, the command line ftp.exe is pretty much straight out of BSD. It is also a single executable that can be removed without breaking the operating system.
I think the stereotype includes the implication that CEO types can't even recognize any increasing trend in tabular form. Sure, graphing almost always makes it easier to spot trends in data, but nobody should be helpless without the graph. (At least for simple matters. I can understand a CEO needing a graph to understand more complex, multivariate data.)
By the way, the most common kinds of graphs found in a spreadsheet app were pretty much all invented by William Playfair, who used line charts, bar charts, pie charts, etc. to study economic issues, not mathematical ones.
Git is designed for use in developing software - specifically, keeping track of source code. The needs of the people actually writing and compiling code takes precedence. That explains why git isn't the greatest tool for delivering software to users.
If you're dealing with big binary files that change frequently, you aren't developing software. You're an artist. You shouldn't be using a tool that's made to assist with things like branches and merges, because those can't be done in an automated fashion for art. Besides, When you're testing software, how often does it really matter if you have the latest-and-greatest textures loaded? (If by chance, you are developing software, you're doing it wrong. Git is for the source code, not the binaries.)
A post saying essentially "eclipse doesn't know how to use git" does not equate to "git sucks balls". If anything, it's a commentary on Eclipse, as it is completely unreasonable to expect the developers of a version control system to create plugins for popular IDEs. I'm sure that if Eclipse's design was such that it was easier to create version control plugins, (perhaps by making it more vcs-agnostic) there would be git plugins on par with what's available for svn, as git clearly exposes all the functionality needed to create such a plugin.
On the other hand, it could simply be a commentary on the types of programmers that use the various tools. Given that Eclipse is a heavily Java-oriented system, and that git was created by Linus for Linux, I think it's clear where you fit into the hierarchy.
From what I've read of release notes for developer seeds (most notably, the current lack of 3d acceleration or power management when using the 64-bit kernel), it sounds like Apple's looking to use 64-bit drivers as much as possible.
And in practice extra x86-64 registers are not that great improvement because modern CPUs got very good at pipelining and data prefetching.
Good point. However, those extra registers may matter quite a bit more for something like the 64-bit Atom processors, which deliberately forgo most speculative features that mitigate register pressure. It would be interesting to see whether it's a better use of silicon to make an out-of-order processor or a 64-bit in-order processor when you're operating under the power constraints of the Atom. The current existence of 64-bit in-order Atom processors suggests that the performance per watt impact of 64-bit is better than out-of-order execution. I suspect this is because 64-bit takes less silicon than OOE, in a similar manner to how useful a good implementation of simultaneous multithreading can be.
Good question, and Apple would be one of the companies to ask first, being that OS X is still 32bit and even the kernel of Snow Leopard will still be 32bit.
[citation needed]
Everything I've seen online (including sources for the wikipedia page on Mac OS 10.6) says that Snow Leopard will have a 64-bit kernel. It also seems that moving to a 64-bit kernel is the most significant 64-bit related improvement for Snow Leopard, as it is the biggest incomplete piece of the 64-bit transition. After that, the only thing left is to compile 64-bit versions of all the bundled apps (which happens to require removing the last vestiges of the Carbon GUI APIs from the Finder).
There wasn't anything wrong with the PowerPC architecture (particularly not age - it's 5 years newer than the 386, and was designed from the start to support 64bit implementations). The fact that all video game consoles use PowerPC or POWER architecture chips shows that there's still life in the architecture. The problem for Apple was the implementations - specifically, the lack of high-performance, low-power chips for laptops.
While it may take a particularly clever compiler to use the extra width of the registers when operating on 32bit data, even the most basic compilers will be able to take advantage of the fact that there are twice as many general purpose registers.
64bit x86 gives you 4 times the general purpose register space and twice as many vector registers, which is a huge benefit for an architecture that's so lacking in register space.
Wrong on several counts. Even treating a modem line as a serial line (which it was), before adding on TCP/IP, the maximum bandwidth supported by the phone system was 56k, due to the bit-robbing scheme used for in-band signaling. In the US, the maximum attainable connection speed was further limited to about 53.3k by FCC limits on the power output of modems. The overhead of PPP, IP, and TCP further subtract from the usable bandwidth.
Try reading that post again. He didn't tout LaTeX as easy to use - he said that Office 2007 makes LaTeX seem easy to use by comparison. That is a commentary on just how bad Office 2007 is.
Also, I call BS on the ribbon being "very well-received" by any group of normal computer users. Any UI change as drastic as the 2003 to 2007 changes will be frustrating for those who have the new version forced upon them, which is almost certainly what happened at the university (seeing as that's what happens in any organization that pays for MS Office site licenses).
I don't think that a 2kbit architecture would be particularly useful. Sure, it might work well for factoring large numbers, but for many other cryptography-related tasks, the lack of granularity would make it rather inefficient. If you're going to build a computer for factoring large numbers, you might as well build TWIRL.
That's right. By default, it downloads updates and then prompts users to install. If a reboot is required, users have the option to defer the installation until the next reboot. There's also always the option to not install the update at all. I agree that Apple's defaults probably wouldn't be a good idea for most Windows users, but they work well enough for me. (And the fact that there's so much outcry over a bad update like this suggests that Mac users are pretty good at patching their systems quickly.)
I've misjudged your bias. You think overclocking is common. It isn't. It definitely isn't among the kinds of people who would switch to a Mac for the sake of stability. Besides, the first step for somebody disappointed with the stability of their system is to stop running the hardware beyond the specs! Crashes caused by hardware failures due to overclocking are not at all germane to a discussion of the relative stability of two software platforms under everyday use, because overclocking is not an everyday task, and if overclocking to a certain extent is going to destabilize your system, it won't matter what operating system you're running.
You're trying to pass off multilingual error messages as worse than stack dumps? Seriously?
At least most people can recognize that the different portions of the OS X kernel panic message have the same meaning, so there's no reason to try to understand those that aren't in your native language. With a windows BSoD, there's way too much information in a language that no regular user is going to be able to parse. So, honestly, which is really worse from a usability standpoint?
Slashdot Mirror
Tufte's work is a great foundation for people who don't know anything about presenting data effectively. I know several people who have been to his seminars, one of whom is a CIO who has since made Tufte's books required reading for most of the people who report to him.
Like the other poster here said, Tufte's stuff isn't that useful for really complex multidimensional data. His stuff is more oriented toward presenting relatively simple stuff in ways that are readable and aren't misleading. On those two subjects, his stuff is very good, and the basic principles all still apply to the more complex stuff the submitter's asking about. The basic principles Tufte presents just don't suffice for high dimensional visualization.
Web sites that require IE are becoming rarer, but they're still around, and usually in the most annoying places. For example, my health insurance provider (which was chosen for me by my employer) has an IE only website. My parents' bank's website requires IE. Several websites I've been required to use during classes at a public university only work fully with IE. If Microsoft had never illegally established their now-eroding web browser monopoly, these websites would have never been developed to require a specific web browser. Firefox hit 1.0 more than 4 years ago, but I still have to keep a Windows virtual machine lying around on my hard drive so that I can use IE. The web definitely hasn't recovered from the harm dealt to it by Microsoft's abuse of monopoly.
The point of all of this is to prevent Microsoft from killing off third party products and companies by bundling free alternatives to the third party products with Windows, thus fixing the price at $0. Microsoft shouldn't have the power to fix prices like that, especially when their abuse of that power bankrupts companies that provided a competitive environment that encourages innovation.
Netscape is the perfect example of why Microsoft's monopoly needs to be policed. After Microsoft killed Netscape, the web browser market stagnated for years, and Microsoft had de facto control over web standards, which resulted in their being no coherent or well-defined standard for building web pages, which enabled Microsoft to create almost as much vendor lock-in as they have with document file formats.
The HTML rendering components should be made into a redistributable package, like the .NET framework. That way, anybody distributing software that needs HTML rendering capabilities can include the installer for the MS HTML engine. This arrangement would also encourage application developers to more carefully consider why and how they add HTML rendering to their app, and whether they should use some other platform, such as XULRunner or Webkit.
That's because ftp.exe was ported almost directly over from BSD, along with the rest of the windows TCP/IP stack. (The internals of the TCP/IP stack have obviously been changed a lot from what they got from BSD, but the utilities like ftp and arp haven't changed much, though Microsoft has buried the Berkeley copyright statements deeper over the years.)
Doesn't WMP include codecs that are covered by patents for which there are licensing fees? If so, the consumer should certainly be able to save money by opting out of licensing the proprietary formats.
There's been plenty of complaining about Microsoft's bundling of a software firewall, though not nearly as much as Microsoft selling security software in a fairly blatant protection racket. The disk defragmenting software was also controversial at one time. The rest of the apps you listed, with the exception of Explorer, can all be completely removed or ignored by the user. (For example, you can easily set it so that all text files open with OO.o instead of Notepad, and then you'll never see Notepad again. You can't do this for all html files and links.)
No, they're arguing that OEMs should be able to pick what browser they want installed. And no, you don't "need" a broswer at install time - most ISPs supply CDs and many of those contain browsers.
... and all of those browsers are very thin wrappers around IE, and can't be modified to use eg. Gecko or WebKit as a back-end.
For me, the big problem with the IE bundling is that many apps (such as email clients) will open html links in IE even if Firefox is the "default" browser. If Microsoft were serious about providing a way to set a default browser, they'd redirect those links to Firefox.
For comparison with another proprietary OS, specifically, Mac OS X: many first-party and built-in apps (help viewer, XCode's doc browser, and Mail.app) use Webkit to display html, but I've never run across an app that, when opening a link in an external browser, insists upon Safari. This may simply be an accident due to the fact that Safari came along long after OS X, but Apple's OS is clearly more accommodating of the user's wishes in this regard.
As I recall, the command line ftp.exe is pretty much straight out of BSD. It is also a single executable that can be removed without breaking the operating system.
I think the stereotype includes the implication that CEO types can't even recognize any increasing trend in tabular form. Sure, graphing almost always makes it easier to spot trends in data, but nobody should be helpless without the graph. (At least for simple matters. I can understand a CEO needing a graph to understand more complex, multivariate data.)
By the way, the most common kinds of graphs found in a spreadsheet app were pretty much all invented by William Playfair, who used line charts, bar charts, pie charts, etc. to study economic issues, not mathematical ones.
And what about the other two relations (voltage and flux, and charge and current)? Are they for some reason not missing?
Git is designed for use in developing software - specifically, keeping track of source code. The needs of the people actually writing and compiling code takes precedence. That explains why git isn't the greatest tool for delivering software to users.
If you're dealing with big binary files that change frequently, you aren't developing software. You're an artist. You shouldn't be using a tool that's made to assist with things like branches and merges, because those can't be done in an automated fashion for art. Besides, When you're testing software, how often does it really matter if you have the latest-and-greatest textures loaded? (If by chance, you are developing software, you're doing it wrong. Git is for the source code, not the binaries.)
A post saying essentially "eclipse doesn't know how to use git" does not equate to "git sucks balls". If anything, it's a commentary on Eclipse, as it is completely unreasonable to expect the developers of a version control system to create plugins for popular IDEs. I'm sure that if Eclipse's design was such that it was easier to create version control plugins, (perhaps by making it more vcs-agnostic) there would be git plugins on par with what's available for svn, as git clearly exposes all the functionality needed to create such a plugin.
On the other hand, it could simply be a commentary on the types of programmers that use the various tools. Given that Eclipse is a heavily Java-oriented system, and that git was created by Linus for Linux, I think it's clear where you fit into the hierarchy.
From what I've read of release notes for developer seeds (most notably, the current lack of 3d acceleration or power management when using the 64-bit kernel), it sounds like Apple's looking to use 64-bit drivers as much as possible.
And in practice extra x86-64 registers are not that great improvement because modern CPUs got very good at pipelining and data prefetching.
Good point. However, those extra registers may matter quite a bit more for something like the 64-bit Atom processors, which deliberately forgo most speculative features that mitigate register pressure. It would be interesting to see whether it's a better use of silicon to make an out-of-order processor or a 64-bit in-order processor when you're operating under the power constraints of the Atom. The current existence of 64-bit in-order Atom processors suggests that the performance per watt impact of 64-bit is better than out-of-order execution. I suspect this is because 64-bit takes less silicon than OOE, in a similar manner to how useful a good implementation of simultaneous multithreading can be.
Good question, and Apple would be one of the companies to ask first, being that OS X is still 32bit and even the kernel of Snow Leopard will still be 32bit.
[citation needed]
Everything I've seen online (including sources for the wikipedia page on Mac OS 10.6) says that Snow Leopard will have a 64-bit kernel. It also seems that moving to a 64-bit kernel is the most significant 64-bit related improvement for Snow Leopard, as it is the biggest incomplete piece of the 64-bit transition. After that, the only thing left is to compile 64-bit versions of all the bundled apps (which happens to require removing the last vestiges of the Carbon GUI APIs from the Finder).
There wasn't anything wrong with the PowerPC architecture (particularly not age - it's 5 years newer than the 386, and was designed from the start to support 64bit implementations). The fact that all video game consoles use PowerPC or POWER architecture chips shows that there's still life in the architecture. The problem for Apple was the implementations - specifically, the lack of high-performance, low-power chips for laptops.
While it may take a particularly clever compiler to use the extra width of the registers when operating on 32bit data, even the most basic compilers will be able to take advantage of the fact that there are twice as many general purpose registers.
64bit x86 gives you 4 times the general purpose register space and twice as many vector registers, which is a huge benefit for an architecture that's so lacking in register space.
Wrong on several counts. Even treating a modem line as a serial line (which it was), before adding on TCP/IP, the maximum bandwidth supported by the phone system was 56k, due to the bit-robbing scheme used for in-band signaling. In the US, the maximum attainable connection speed was further limited to about 53.3k by FCC limits on the power output of modems. The overhead of PPP, IP, and TCP further subtract from the usable bandwidth.
Try reading that post again. He didn't tout LaTeX as easy to use - he said that Office 2007 makes LaTeX seem easy to use by comparison. That is a commentary on just how bad Office 2007 is.
Also, I call BS on the ribbon being "very well-received" by any group of normal computer users. Any UI change as drastic as the 2003 to 2007 changes will be frustrating for those who have the new version forced upon them, which is almost certainly what happened at the university (seeing as that's what happens in any organization that pays for MS Office site licenses).
I don't think that a 2kbit architecture would be particularly useful. Sure, it might work well for factoring large numbers, but for many other cryptography-related tasks, the lack of granularity would make it rather inefficient. If you're going to build a computer for factoring large numbers, you might as well build TWIRL.
That's right. By default, it downloads updates and then prompts users to install. If a reboot is required, users have the option to defer the installation until the next reboot. There's also always the option to not install the update at all. I agree that Apple's defaults probably wouldn't be a good idea for most Windows users, but they work well enough for me. (And the fact that there's so much outcry over a bad update like this suggests that Mac users are pretty good at patching their systems quickly.)
I've misjudged your bias. You think overclocking is common. It isn't. It definitely isn't among the kinds of people who would switch to a Mac for the sake of stability. Besides, the first step for somebody disappointed with the stability of their system is to stop running the hardware beyond the specs! Crashes caused by hardware failures due to overclocking are not at all germane to a discussion of the relative stability of two software platforms under everyday use, because overclocking is not an everyday task, and if overclocking to a certain extent is going to destabilize your system, it won't matter what operating system you're running.
You're trying to pass off multilingual error messages as worse than stack dumps? Seriously?
At least most people can recognize that the different portions of the OS X kernel panic message have the same meaning, so there's no reason to try to understand those that aren't in your native language. With a windows BSoD, there's way too much information in a language that no regular user is going to be able to parse. So, honestly, which is really worse from a usability standpoint?