I'll admit, MS desktop has *very* low latency at idle, and runs circles around Linux camp (with all those linked libraries linking all over the place) at starting up applications.
So does anybody know offhand what the average number of DLL's used by typical Windows apps is, and the average number of.so's used by typical, for example, KDE or GNOME apps is? (You could probably beat Windows by running lighter-weight applications, but comparing with KDE or GNOME is probably "fairer".) Do Windows apps really have fewer linked libraries than KDE/GNOME apps on the average?
Not for the standard GUI, no: that's handled by the rendering being done by libraries in the application process and then handed to Quartz Compositor to be drawn on the screen, as per this overview from Apple.
X11.app is one of the clients of Quartz Compositor, and used only when you run X applications and have them display on the machine running OS X (whether they're running there or not).
From his post, the only size-related thing is its flash vs. the hard drive of the iPhone.
That would make it SMALLER.
Except that the iPhone doesn't have a hard drive; it has flash. The specs for the HTC P4550 say it's 59x112x19 mm, 190 g with battery. The specs for the iPhone say it's 61x115x11.6mm, so the iPhone is a little longer and wider and a little thinner, and it weighs 135 g (and the battery's attached to the motherboard, so presumably that's "with battery"), so the iPhone is a little lighter.
Mediawiki has already added the capability to look at the Special:Contributions for an IP range. I'm not sure if it's been enabled yet on EN.
If you click on the IP address in an anonymous change in a history, it takes you to a list of that IP address's changes. The URL it takes you to is http://en.wikipedia.org/wiki/Special:Contributions /IP-address, where "IP-address" is the dotted-quad form of the IP address.
As I remember it. Altos was hired by USL to merge Unix System. Sun was hired to merge Berkeley code into Unix. Finally Sun was hired to merge the resulting code and that became SVR4. USL was only a management front and did no coding. Novel bought that management organization.
As I remember it, from when I was in the OS group at Sun (up to late 1988), Altos wasn't involved at all, and Sun and the Unix development group at AT&T (in Summit, New Jersey) were the ones who did the coding. USL was a later spinoff of the latter group as a wholly-owned AT&T subsidiary, later bought by Novell; it was not a "management front".
No one I met in the field called SunOS + Openwindows Solaris, until Solaris 2.x came out.
That's because Sun didn't introduce the "Solaris" brand until SunOS 5.0 came out, at which point they started using it for the 4.1[.x]+window system combination ("Solaris 1.x") and the 5.x+window system combination ("Solaris 2.x").
Broadband is somewhat cheaper in Holland than it is here, but rest assured that it won't be cheaper than the same in the US.
Well, KPN says that they offer (at least from what this non-Dutch-speaker can pick out with the aid of Babelfish) 1.5Mb down/256Kb up for EUR 15.00/month for the first year and EUR 25.00/month after that, 3Mb down/512Kb up for EUR 20.00/EUR 30.00, 6Mb down/768 Kb up for EUR 40.00/EUR 50.00. The "packages" (Internet + TV + telephone) at those speeds go for EUR 44.90/month and EUR 52.90/month - unless I'm missing something, they don't have a 6Mb up/768 Kb down package.
Here in Amurrica, at least in California, AT&T says that they offer 768Kb down/128Kb up for USD 14.99/month, "up to" 3MB down/"up to" 512Kb down for USD 24.99/month, and "up to" 6MB down/"up to" 768Kb down for USD 34.99/month. They have an Internet + TV + long-distance (you have to get your local phone service from them to get DSL) bundle for USD 99.98/month - as the Internet part is "Internet Pro", and "Pro" for their "just DSL" offers is the 3MB down/512Kb up service, I assume that's what you get for the DSL with the bundle.
I don't know whether "up to" means that they're just being more cautious in what they state than KPN, or if KPN really does offer guarantees that AT&T doesn't. I don't see anything in the AT&T material I saw about the rate going up after the first year.
On the other hand, perhaps the Dutch and Belgians should move to France. Orange appears to offer 1Mb down/??? up for EUR 24.90/month, up to 8 Mb down/800 Kb up for EUR 29.90/month, and up to 18 Mb down/800Kb up for EUR 34.90/month - and, unless I'm misreading (which this non-French-speaker might be doing), that includes TV. It also appears that they offer an Internet+TV+telephone bundle, with an Mb down/800 Kb up, for EUR 34.90/month, with a 1-year commitment (it sounds as if the "a la carte" requires no commitment).
I didn't see anything on KPN, AT&T, or Orange about a traffic limit.
At the time, Sun claimed that Solaris 1.x was SunOS with Openview
They did so because it was. Back when the SVR4 project started, if there was any notion of using the term "Solaris", nobody told us in Engineering about it; as far as we knew, the new OS was going to be called SunOS 5.x.
If I remember correctly, when Sun moved from Sun/OS to Solaris (and from kind of mix of BSD and SysV to pure SysV), first they didn't want to get a SysV license
You do not remember correctly. Sun most definitely got an SV license at least as far back as SunOS 3.2, which picked up a large chunk of System V code (yes, even before SunOS 4.0). I don't remember what UNIX license they had earlier, but even in 3.0 there was, as I remember, some SV code (I think the SunOS 3.0 Bourne shell was an SVR2 Bourne shell tweaked to be more V7/BSD-compatible and the SunOS 3.0 Berkmail was SVR2 mailx - itself based on Berkmail - tweaked similarly).
The SVR4 project whence SunOS 5.x came was an AT&T/Sun joint project, and Sun hardly would have wanted to avoid getting an SVR4 license. It was most definitely based on AT&T code - although a lot of the "AT&T code" in SVR4 was, in turn, based on Sun code (e.g., SVR4's VM system was derived from SunOS 4.x's).
(Oh, and Solaris 1 was based on SunOS 4.1[.x]; Solaris 2 was the name for the Solaris that used the SVR4-based SunOS 5.x, and, until SunOS 5.7, Solaris 2.x had SunOS 5.x as its core OS - eventually, I guess Sun decided that "Solaris 3" in the sense of a complete rewrite with the OS becoming "SunOS 6.0" wasn't going to happen any time soon, so they got rid of the no-longer-very-interesting "2." and just went to "Solaris 7", followed by Solaris 8, 9, and 10.
In addition, the license purchase didn't have anything to do with the "start of a stable Solaris" - that was, from everything I know, the result of a lot of people at Sun doing a lot of work on the OS to beat it into shape.)
Nothing about POSIX requires that it be impossible to get higher-resolution time stamps. You just can't get them with POSIX-only code.
For instance, OS X (and possibly other BSD-flavored UN*Xes) either defines the time stamps, in "struct stat", as struct timespec st_[amc]timespec; and #defines st_[amc]time as st_[amc]timespec.tv_sec, or puts a long st_[amc]timensec; field after each time_t st_[amc]time; field, depending on whether _POSIX_C_SOURCE is defined - POSIX-only code will work, and non-POSIX-only code can get at the higher-resolution times.
And the POSIX requirement for st_atime to be updated has nothing to do with the issue you're complaining about in any case.
How many of the groups tests for cerification if any, are totally or partially political or financial or contractual in nature ?
Few, if any, as far as I know; the tests are a large pile of code you run on your system, and you pass or fail. There's a licensing fee for the UNIX trademark, and you presumably end up signing a contract to license the trademark.
Why isn't OS X just linux?
Because Linux didn't exist at the time NeXT was founded, OS X is a NEXTSTEP descendant, and it presumably wasn't considered worth the effort to construct OS X and all the frameworks in it atop Linux.
Wont many Linux commands typed into oS x run fine?
Yes, and many Linux commands typed into Solaris will run fine, and many Linux commands typed into AIX will run fine, and many Linux commands typed into {Free,Net,Open,DragonFly}BSD will run fine, and so on. In some cases, that's because the Linux command in question was designed so that {Solaris,AIX,etc.} commands typed into Linux would run fine.
Of what benefit is this to Uses?
It depends on the user. If the user is somebody who has code that expects Single UNIX Standard behavior (e.g., that all the thread-cancellation stuff works), it means their code should work on OS X. If the user is somebody who wants to use an application with that code, it means there might be a better chance that said application will be ported to OS X (although, if the app is a GUI app, it'll either run only with the X server and won't look particularly native, or would have to be ported, or would have to be written with a cross-platform toolkit such as Qt).
Apple clearly uses their BSD subsystem as a license-free compatibility layer- it's clear to me that proper POSIX compliance was not the most important factor in the system design. Then again, Linux isn't really POSIX compliant either- but that's another tale for another time.
And note that said "license-free compatibility layer" is the layer at which, for example, file I/O and network I/O are implemented (except for the driver layer, which is done with I/O Kit, if you consider I/O Kit separate from "BSD" - it's in a separate subtree of the XNU source, along with the Mach code). It's not as if Carbon and Cocoa and Core Foundation are using their own separate calls to do file and network operations.
Then consider that Apple has also moved architectures (PPC -> x86) since the introduction of OS X, and probably will again (x86 -> AMD64 - they ship that hardware, but the OS is still at least mostly x86).
Leopard will, at least, be supporting both 32-bit and 64-bit versions of most system libraries and frameworks, which is less "moving from x86 to AMD64/Intel 64/x86-64/whatever" than "more fully supporting both PPC64 and AMD64/Intel 64/x86-64/whatever" - it's not as if Leopard will support only 64-bit binaries.
Yes, most of it is derived from FreeBSD's libc, although not all the stuff comes from FreeBSD (the memory allocator is Apple's own, and the get*by* routines talk to a daemon via Mach messages rather than going through a name service switch to routines that directly read files or talk to servers, for example). It is most definitely NOT GNU libc (heck, it's not even called "libc", it's called "libSystem" - but it has the same role that the library called "libc" on other UN*Xes has).
This response to the article explains it. (Executive summary: just because Microsoft stopped selling NT as "Windows NT" followed by a version number, that doesn't mean the version numbers went away.)
do you know how it translates microcode on the software level to simplify hardware etc?
No, I don't. I do, however, know that they had software that ran on the chip that translated x86 machine code (not microcode) to native VLIW code and ran the resulting code.
They do apparently indeed patch the microcode. It is direct equivalent to an IBM CPU, and is not an emulation. Originally they did higher-level emulation, but they are apparently now mucking with the microcode and run the opcodes directly on the CPU. That is, they turn the Itanium into a z/CPU.
Then they probably got Intel to put hardware into the Itanium 2 to help them - modern CPUs tend to do the instruction decoding in hardware, and, if they use microcode, hand off sme instructions to microcode (but not all instructions - even x86 chips run the most common instructions directly in hardware). z/Architecture instructions look very different from Itanium instructions, so the instruction decoding hardware on Itanium 2 is likely to be of little if any use in decoding z/Architecture instructions.
But yes, from what I'm told PSI does indeed do microcode.
Including the Itaniums? Note that those are the processors that Platform Solutions, Inc. uses. They might have some microcode in the part of the chip that runs legacy x86 code, but that part is, from what I've heard, not very fast. I don't know whether they have any microcode involved in running Itanium instructions, however.
Slashdot Mirror
Dave's one of the founders, but he's not the CEO or president, he's an executive vice president, as per the NetApp executive biographies page.
a = dv/dt, so F = m dv/dt. Calculus-based enough for you?
What, you don't believe this?
So does anybody know offhand what the average number of DLL's used by typical Windows apps is, and the average number of .so's used by typical, for example, KDE or GNOME apps is? (You could probably beat Windows by running lighter-weight applications, but comparing with KDE or GNOME is probably "fairer".) Do Windows apps really have fewer linked libraries than KDE/GNOME apps on the average?
Not for the standard GUI, no: that's handled by the rendering being done by libraries in the application process and then handed to Quartz Compositor to be drawn on the screen, as per this overview from Apple.
X11.app is one of the clients of Quartz Compositor, and used only when you run X applications and have them display on the machine running OS X (whether they're running there or not).
And a lot of Iranians have probably thought of us (and Britain) as an enemy in one way or another since this happened.
Except that the iPhone doesn't have a hard drive; it has flash. The specs for the HTC P4550 say it's 59x112x19 mm, 190 g with battery. The specs for the iPhone say it's 61x115x11.6mm, so the iPhone is a little longer and wider and a little thinner, and it weighs 135 g (and the battery's attached to the motherboard, so presumably that's "with battery"), so the iPhone is a little lighter.
Neither of which the iPhone has (its file system is in flash memory).
If you click on the IP address in an anonymous change in a history, it takes you to a list of that IP address's changes. The URL it takes you to is http://en.wikipedia.org/wiki/Special:Contributions /IP-address, where "IP-address" is the dotted-quad form of the IP address.
As I remember it, from when I was in the OS group at Sun (up to late 1988), Altos wasn't involved at all, and Sun and the Unix development group at AT&T (in Summit, New Jersey) were the ones who did the coding. USL was a later spinoff of the latter group as a wholly-owned AT&T subsidiary, later bought by Novell; it was not a "management front".
That's because Sun didn't introduce the "Solaris" brand until SunOS 5.0 came out, at which point they started using it for the 4.1[.x]+window system combination ("Solaris 1.x") and the 5.x+window system combination ("Solaris 2.x").
Well, KPN says that they offer (at least from what this non-Dutch-speaker can pick out with the aid of Babelfish) 1.5Mb down/256Kb up for EUR 15.00/month for the first year and EUR 25.00/month after that, 3Mb down/512Kb up for EUR 20.00/EUR 30.00, 6Mb down/768 Kb up for EUR 40.00/EUR 50.00. The "packages" (Internet + TV + telephone) at those speeds go for EUR 44.90/month and EUR 52.90/month - unless I'm missing something, they don't have a 6Mb up/768 Kb down package.
Here in Amurrica, at least in California, AT&T says that they offer 768Kb down/128Kb up for USD 14.99/month, "up to" 3MB down/"up to" 512Kb down for USD 24.99/month, and "up to" 6MB down/"up to" 768Kb down for USD 34.99/month. They have an Internet + TV + long-distance (you have to get your local phone service from them to get DSL) bundle for USD 99.98/month - as the Internet part is "Internet Pro", and "Pro" for their "just DSL" offers is the 3MB down/512Kb up service, I assume that's what you get for the DSL with the bundle.
I don't know whether "up to" means that they're just being more cautious in what they state than KPN, or if KPN really does offer guarantees that AT&T doesn't. I don't see anything in the AT&T material I saw about the rate going up after the first year.
On the other hand, perhaps the Dutch and Belgians should move to France. Orange appears to offer 1Mb down/??? up for EUR 24.90/month, up to 8 Mb down/800 Kb up for EUR 29.90/month, and up to 18 Mb down/800Kb up for EUR 34.90/month - and, unless I'm misreading (which this non-French-speaker might be doing), that includes TV. It also appears that they offer an Internet+TV+telephone bundle, with an Mb down/800 Kb up, for EUR 34.90/month, with a 1-year commitment (it sounds as if the "a la carte" requires no commitment).
I didn't see anything on KPN, AT&T, or Orange about a traffic limit.
They did so because it was. Back when the SVR4 project started, if there was any notion of using the term "Solaris", nobody told us in Engineering about it; as far as we knew, the new OS was going to be called SunOS 5.x.
You do not remember correctly. Sun most definitely got an SV license at least as far back as SunOS 3.2, which picked up a large chunk of System V code (yes, even before SunOS 4.0). I don't remember what UNIX license they had earlier, but even in 3.0 there was, as I remember, some SV code (I think the SunOS 3.0 Bourne shell was an SVR2 Bourne shell tweaked to be more V7/BSD-compatible and the SunOS 3.0 Berkmail was SVR2 mailx - itself based on Berkmail - tweaked similarly).
The SVR4 project whence SunOS 5.x came was an AT&T/Sun joint project, and Sun hardly would have wanted to avoid getting an SVR4 license. It was most definitely based on AT&T code - although a lot of the "AT&T code" in SVR4 was, in turn, based on Sun code (e.g., SVR4's VM system was derived from SunOS 4.x's).
(Oh, and Solaris 1 was based on SunOS 4.1[.x]; Solaris 2 was the name for the Solaris that used the SVR4-based SunOS 5.x, and, until SunOS 5.7, Solaris 2.x had SunOS 5.x as its core OS - eventually, I guess Sun decided that "Solaris 3" in the sense of a complete rewrite with the OS becoming "SunOS 6.0" wasn't going to happen any time soon, so they got rid of the no-longer-very-interesting "2." and just went to "Solaris 7", followed by Solaris 8, 9, and 10.
In addition, the license purchase didn't have anything to do with the "start of a stable Solaris" - that was, from everything I know, the result of a lot of people at Sun doing a lot of work on the OS to beat it into shape.)
Nothing about POSIX requires that it be impossible to get higher-resolution time stamps. You just can't get them with POSIX-only code.
For instance, OS X (and possibly other BSD-flavored UN*Xes) either defines the time stamps, in "struct stat", as struct timespec st_[amc]timespec; and #defines st_[amc]time as st_[amc]timespec.tv_sec, or puts a long st_[amc]timensec; field after each time_t st_[amc]time; field, depending on whether _POSIX_C_SOURCE is defined - POSIX-only code will work, and non-POSIX-only code can get at the higher-resolution times.
And the POSIX requirement for st_atime to be updated has nothing to do with the issue you're complaining about in any case.
What would you use at an ATM machine other than a PIN number?
Few, if any, as far as I know; the tests are a large pile of code you run on your system, and you pass or fail. There's a licensing fee for the UNIX trademark, and you presumably end up signing a contract to license the trademark.
Because Linux didn't exist at the time NeXT was founded, OS X is a NEXTSTEP descendant, and it presumably wasn't considered worth the effort to construct OS X and all the frameworks in it atop Linux.
Yes, and many Linux commands typed into Solaris will run fine, and many Linux commands typed into AIX will run fine, and many Linux commands typed into {Free,Net,Open,DragonFly}BSD will run fine, and so on. In some cases, that's because the Linux command in question was designed so that {Solaris,AIX,etc.} commands typed into Linux would run fine.
It depends on the user. If the user is somebody who has code that expects Single UNIX Standard behavior (e.g., that all the thread-cancellation stuff works), it means their code should work on OS X. If the user is somebody who wants to use an application with that code, it means there might be a better chance that said application will be ported to OS X (although, if the app is a GUI app, it'll either run only with the X server and won't look particularly native, or would have to be ported, or would have to be written with a cross-platform toolkit such as Qt).
UNIX certification.
As opposed to OS X, where not just POSIX compliance, but Single UNIX Standard v3 compliance is a tale for this October.
And note that said "license-free compatibility layer" is the layer at which, for example, file I/O and network I/O are implemented (except for the driver layer, which is done with I/O Kit, if you consider I/O Kit separate from "BSD" - it's in a separate subtree of the XNU source, along with the Mach code). It's not as if Carbon and Cocoa and Core Foundation are using their own separate calls to do file and network operations.
Leopard will, at least, be supporting both 32-bit and 64-bit versions of most system libraries and frameworks, which is less "moving from x86 to AMD64/Intel 64/x86-64/whatever" than "more fully supporting both PPC64 and AMD64/Intel 64/x86-64/whatever" - it's not as if Leopard will support only 64-bit binaries.
Yes, most of it is derived from FreeBSD's libc, although not all the stuff comes from FreeBSD (the memory allocator is Apple's own, and the get*by* routines talk to a daemon via Mach messages rather than going through a name service switch to routines that directly read files or talk to servers, for example). It is most definitely NOT GNU libc (heck, it's not even called "libc", it's called "libSystem" - but it has the same role that the library called "libc" on other UN*Xes has).
What bands were you thinking of in addition to 850, 900, 1800, and 1900 MHz?
This response to the article explains it. (Executive summary: just because Microsoft stopped selling NT as "Windows NT" followed by a version number, that doesn't mean the version numbers went away.)
No, I don't. I do, however, know that they had software that ran on the chip that translated x86 machine code (not microcode) to native VLIW code and ran the resulting code.
Then they probably got Intel to put hardware into the Itanium 2 to help them - modern CPUs tend to do the instruction decoding in hardware, and, if they use microcode, hand off sme instructions to microcode (but not all instructions - even x86 chips run the most common instructions directly in hardware). z/Architecture instructions look very different from Itanium instructions, so the instruction decoding hardware on Itanium 2 is likely to be of little if any use in decoding z/Architecture instructions.
What precisely were you told?
Including the Itaniums? Note that those are the processors that Platform Solutions, Inc. uses. They might have some microcode in the part of the chip that runs legacy x86 code, but that part is, from what I've heard, not very fast. I don't know whether they have any microcode involved in running Itanium instructions, however.