US public schools are not, in general, poorly funded. See this report, particularly the international comparison of spending per student in table 4. The report concludes that "What distinguishes U.S. education from other nations is not how much or how little it spends. Thousands of schools district make the U.S. unique. Education costs and personnel compensation vary tremendously. School districts operate free of national education standards and national salary schedules. The U.S. has forged a structure of long school days, busing, and school lunch programs to support a system of large schools. Consequently, noninstructional spending is a larger portion of school costs."
Imagine how cool it would be to have every cell provider in the US use the same towers!
There are actually shared towers in some places in the UK, though they aren't the norm. Not every network needs cells in the same places, as they use different protocols and/or frequencies (hence different propagation characteristics) and have different customer bases (hence different call density distribution).
Imagine being able to separate your cable TV connection from your cable TV content provider.
I just want to pick and choose channels.
Imagine being able to buy power from your favorite [green source du jour] - even across the country.
Here in the UK you can choose your electricity and natural gas suppliers. They provide customer service and billing. Obviously the electricity and gas continues to be physically supplied by the local utility, but the chosen "supplier" pays producers to put it into the network. I don't know quite how that works in the absence of real-time metering.
Yuck. Gimp (1.2) doesn't seem to use the X clipboard. Pasting a web page fragment with images from Mozilla into OOWriter causes it to hang (it appears to copy as HTML and then attempt to download the images, but somehow fails to do so). I don't know what other apps to try this with.
You seem to be describing isapnptools here (1. Run pnpdump. 2. Edit/etc/isapnp.conf.) This is indeed annoyingly complex, but it's just the same as what you get in the Windows Device Manager for ISA PnP devices.
Now I have no idea why you need to mess with PnP just to use a serial port.
As each disk's filesystem has a volume ID as well as a label, the Amiga could tell if you inserted the wrong disk with the same name, and would put the requester up again and again until you inserted the right disk.
The "volume ID" was the creation timestamp. Usual practice was to increment the creation timestamp by one tick (20 ms) when duplicating a disk. OS 1.x would crash if it found two disks present with the same label and timestamp!
It wasn't really mounting/unmounting. The OS makes a distinction between mounting a device (starting a file-handler, or associating a file-system with a low-level disk device) and getting an actual volume (disk) in that device. Devices normally remain mounted forever.
What I have described is the way AmigaOS handles floppies (and probably also CDs, but I never owned an Amiga with CDROM). And it is the way I want Linux to handle CDs, and other removable media as far as hardware permits. Of course in Linux it gets complicated by X vs. terminals and the fact that the graphical user interface is not a part of the kernel. It also gets complicated by the fact there can be multiple users. But I have ideas about how to handle all of those problems.
I think AmigaOS already indicates the way to go. It had a field in the process structure (pr_WindowPtr) which indicated where requests should go when the process made a file-system call that encountered a user-correctable error such as the disk having been removed. If it was set to -1 the call would always return an error immediately. If it was set to 0 the OS would open a request on the default screen. Otherwise it was a pointer to a window, and the OS would open a request on the same screen as that window.
So in Unix there could be a sumilar field, but it would indicate a socket or some other IPC mechanism to send messages through in such circumstances. If clear, the call would return an error as usual. If set, the thread would be blocked and the kernel would send a message describing the error. When it received an appropriate message back it would unblock the thread and retry or fail.
An X session manager could start a requester daemon to handle such messages and set this field in all the other child processes it spawns.
There is a question as to whether and when the field should be cleared automatically, bearing in mind that we have a large legacy of software that doesn't know about it. It shouldn't be inherited by a daemon that's started from a shell in an xterm in X, for example. Perhaps setsid() should clear it.
The Amiga used floppy drives that had different hardware than the PC, so maybe that feature was part of the Amiga hardware.
Every PC floppy drive has a "disk removed" output line (pin 34, DSKCHG). The Amiga has a different FD controller and some additional logic for each drive, but it uses the same drives (except that high-density drives require some kluges as the FD controller can't handle the high-density data rate).
However, I seem to remember that when you took a disk out of an Amiga drive, you'd hear a periodic soft click, like maybe every few seconds. Perhaps that was sort of a 'ping' that looked to see if the disk was present or not, or if it had changed.
Right. The problem is that the floppy drive bus can be shared between 2 or more drives, so the DSKCHG line is controlled by the currently selected drive and if a disk is changed in a drive that isn't currently selected this may not be noticed. The way the Amiga dealt with this was to have some additional logic between each drive and the controller. All drives see themselves as currently selected, and set their output lines accordingly. The DSKCHG signal is latched if it is activated. When the drive is really selected, the driver can detect a disk removal that happened while it was deselected. The protocol for clearing the latch, to find out whether a disk has been inserted, is to step the drive heads. This is what causes the click.
There is a way to avoid the click, which is to send a step-out signal when the heads are already on cylinder 0 (the outermost cylinder). Most floppy drives will ignore this. Unfortunately some are not so smart and can be damaged by this, which is why it is not done by default. A better approach would be to hide the latch-clearing step signal from the drive, but that was presumably considered unreasonably complex.
It may have worked that way in 16-bit Windows, but in Win32 each process has its own set of DLLs. This problem shouldn't occur in Win32 - unless a single program indirectly depends on two different DLLs with the same base name.
The Amiga had a fairly decent CLI, but in early versions of the OS there was an option to hide the CLI icon in Workbench (the Amiga's Explorer/Finder equivalent). I have a suspicion that that was the default. What's worse, there was no way to run a single command from Workbench in those early versions!
Here in the UK, the first in the pecking order is the government claiming taxes. I wouldn't be surprised if other governments reserved the same privilege.
That patent is gobbledegook! At some point someone has gone through it and done a global search and replace of normal words into patentese:
s/interface/one interface/g
s/interfaces/at least one interface/g
s/\(originating\|destination\) processor/one of the plurality of \1 processors/g
s/\(originating\|destination\) processors/at least one of the plurality of \1 processors/g
The result is ungrammatical in many places ("...the at least one...". I think that should be reason enough to invalidate it. Can anyone submit such nonsense in good faith?
So Microsoft is going to start making microwaves? (Consumer electronics manufacturers would never go into partnership with them.) It sounds like a stupid marketing exercise to me. Anyway, since I have to go up to the microwave to put food in, what benefit would I get from using a remote control?
Microsoft's university partners also exhibited projects. Brad Myers, a research scientist at Carnegie Mellon University, has been working to replace the remote controls lying around the home with one device, such as a cellphone or a personal digital assistant. Eventually, he said, appliances could be equipped with technology to receive the commands.
Which rock has he been living under? Combined remotes have been around for, what, 10 years or so, and remote control software is readily available for PDAs with IR ports.
Re:It's not even a matter of checking user input!
on
LSB & Posix Conflicts
·
· Score: 1
I searched for and eventually found some relevant history. It turns out that gets() was part of the "portable C library" written by Mike Lesk and released with version 6 of Unix. fgets() is part of stdio, which was written by Dennis Ritchie and released with version 7. Most of the "portable C library" seems to have been changed or discarded, but gets() lives on.
Re:It's not even a matter of checking user input!
on
LSB & Posix Conflicts
·
· Score: 1
No, the article says gets() is not part of the string library (in which some functions have length-limited variants). gets() is not mentioned in K&R first edition (1978), whereas fgets() is. I would be inclined to blame Berkeley for the fuzzy thinking that must be behind gets().
Yup, this seems to be the way to do it. I still feel uneasy about that, though. Remember when Windows Update got hit by Code Red? Obviously those web servers aren't always kept up to date themselves, so maybe someone could replace some of the patches on them. Besides which, Windows Update is not at all reliable.
Of course, you have know way of ensuring that the certificate isn't tampered with before or during the download. (You could send some kind of digest by a secure channel, but then I don't think IE shows digests.)
The flat fee is based on an archaic property valuation - the 1973 rateable value, reflecting market rents that year. If you live in a flat (highly desirable back in 1973, for some reason) or a large house it's generally better to get a meter.
Slashdot Mirror
US public schools are not, in general, poorly funded. See this report, particularly the international comparison of spending per student in table 4. The report concludes that "What distinguishes U.S. education from other nations is not how much or how little it spends. Thousands of schools district make the U.S. unique. Education costs and personnel compensation vary tremendously. School districts operate free of national education standards and national salary schedules. The U.S. has forged a structure of long school days, busing, and school lunch programs to support a system of large schools. Consequently, noninstructional spending is a larger portion of school costs."
He never claimed either of those, but certain people wanted to make him out to be a liar.
There are actually shared towers in some places in the UK, though they aren't the norm. Not every network needs cells in the same places, as they use different protocols and/or frequencies (hence different propagation characteristics) and have different customer bases (hence different call density distribution).
I just want to pick and choose channels.
Here in the UK you can choose your electricity and natural gas suppliers. They provide customer service and billing. Obviously the electricity and gas continues to be physically supplied by the local utility, but the chosen "supplier" pays producers to put it into the network. I don't know quite how that works in the absence of real-time metering.
Yuck. Gimp (1.2) doesn't seem to use the X clipboard. Pasting a web page fragment with images from Mozilla into OOWriter causes it to hang (it appears to copy as HTML and then attempt to download the images, but somehow fails to do so). I don't know what other apps to try this with.
No, it's not free. Red Hat and other distributors are charging big money, and they need to deliver something that's worth it.
If you advocate Linux, you should be prepared to address the problems that people have with it.
I tried copy-and-pasting a table from Mozilla into OOCalc, and it just worked. I'd say it's up to speed, even if "IAmTheRealMike" didn't know that.
You seem to be describing isapnptools here (1. Run pnpdump. 2. Edit /etc/isapnp.conf.) This is indeed annoyingly complex, but it's just the same as what you get in the Windows Device Manager for ISA PnP devices.
Now I have no idea why you need to mess with PnP just to use a serial port.
The "volume ID" was the creation timestamp. Usual practice was to increment the creation timestamp by one tick (20 ms) when duplicating a disk. OS 1.x would crash if it found two disks present with the same label and timestamp!
It wasn't really mounting/unmounting. The OS makes a distinction between mounting a device (starting a file-handler, or associating a file-system with a low-level disk device) and getting an actual volume (disk) in that device. Devices normally remain mounted forever.
I think AmigaOS already indicates the way to go. It had a field in the process structure (pr_WindowPtr) which indicated where requests should go when the process made a file-system call that encountered a user-correctable error such as the disk having been removed. If it was set to -1 the call would always return an error immediately. If it was set to 0 the OS would open a request on the default screen. Otherwise it was a pointer to a window, and the OS would open a request on the same screen as that window.
So in Unix there could be a sumilar field, but it would indicate a socket or some other IPC mechanism to send messages through in such circumstances. If clear, the call would return an error as usual. If set, the thread would be blocked and the kernel would send a message describing the error. When it received an appropriate message back it would unblock the thread and retry or fail.
An X session manager could start a requester daemon to handle such messages and set this field in all the other child processes it spawns.
There is a question as to whether and when the field should be cleared automatically, bearing in mind that we have a large legacy of software that doesn't know about it. It shouldn't be inherited by a daemon that's started from a shell in an xterm in X, for example. Perhaps setsid() should clear it.
Every PC floppy drive has a "disk removed" output line (pin 34, DSKCHG). The Amiga has a different FD controller and some additional logic for each drive, but it uses the same drives (except that high-density drives require some kluges as the FD controller can't handle the high-density data rate).
Right. The problem is that the floppy drive bus can be shared between 2 or more drives, so the DSKCHG line is controlled by the currently selected drive and if a disk is changed in a drive that isn't currently selected this may not be noticed. The way the Amiga dealt with this was to have some additional logic between each drive and the controller. All drives see themselves as currently selected, and set their output lines accordingly. The DSKCHG signal is latched if it is activated. When the drive is really selected, the driver can detect a disk removal that happened while it was deselected. The protocol for clearing the latch, to find out whether a disk has been inserted, is to step the drive heads. This is what causes the click.
There is a way to avoid the click, which is to send a step-out signal when the heads are already on cylinder 0 (the outermost cylinder). Most floppy drives will ignore this. Unfortunately some are not so smart and can be damaged by this, which is why it is not done by default. A better approach would be to hide the latch-clearing step signal from the drive, but that was presumably considered unreasonably complex.
Unfortunately apt ignores Recommends and Suggests, because it's not interactive.
It may have worked that way in 16-bit Windows, but in Win32 each process has its own set of DLLs. This problem shouldn't occur in Win32 - unless a single program indirectly depends on two different DLLs with the same base name.
And as we all know, Windows installers are consistent, reliable, and compact. Right.
The Amiga had a fairly decent CLI, but in early versions of the OS there was an option to hide the CLI icon in Workbench (the Amiga's Explorer/Finder equivalent). I have a suspicion that that was the default. What's worse, there was no way to run a single command from Workbench in those early versions!
Here in the UK, the first in the pecking order is the government claiming taxes. I wouldn't be surprised if other governments reserved the same privilege.
- s/interface/one interface/g
- s/interfaces/at least one interface/g
- s/\(originating\|destination\) processor/one of the plurality of \1 processors/g
- s/\(originating\|destination\) processors/at least one of the plurality of \1 processors/g
The result is ungrammatical in many places ("...the at least one...". I think that should be reason enough to invalidate it. Can anyone submit such nonsense in good faith?So Microsoft is going to start making microwaves? (Consumer electronics manufacturers would never go into partnership with them.) It sounds like a stupid marketing exercise to me. Anyway, since I have to go up to the microwave to put food in, what benefit would I get from using a remote control?
Which rock has he been living under? Combined remotes have been around for, what, 10 years or so, and remote control software is readily available for PDAs with IR ports.
I searched for and eventually found some relevant history. It turns out that gets() was part of the "portable C library" written by Mike Lesk and released with version 6 of Unix. fgets() is part of stdio, which was written by Dennis Ritchie and released with version 7. Most of the "portable C library" seems to have been changed or discarded, but gets() lives on.
No, the article says gets() is not part of the string library (in which some functions have length-limited variants). gets() is not mentioned in K&R first edition (1978), whereas fgets() is. I would be inclined to blame Berkeley for the fuzzy thinking that must be behind gets().
"You can get this for 10 cents less at Aldi." (message charge: 20 cents)
Yup, this seems to be the way to do it. I still feel uneasy about that, though. Remember when Windows Update got hit by Code Red? Obviously those web servers aren't always kept up to date themselves, so maybe someone could replace some of the patches on them. Besides which, Windows Update is not at all reliable.
Of course, you have know way of ensuring that the certificate isn't tampered with before or during the download. (You could send some kind of digest by a secure channel, but then I don't think IE shows digests.)
The flat fee is based on an archaic property valuation - the 1973 rateable value, reflecting market rents that year. If you live in a flat (highly desirable back in 1973, for some reason) or a large house it's generally better to get a meter.