The focus of the article is that Microsoft is ignoring people who cry "privacy" and opt for updates by default with a choice to disable or screen because they cannot allow Slammer worms to thrive when patches have been out for months.
Also, they are going to ship with firewalls enabled. People who want to run servers will have to learn to open ports.
Microsoft is adopting a secure-by-default stance, even if it inconviences some users or ruffles a few feathers.
The fact that Debian can do these things has nothing to do with it. I take it you are very proud of Debian, but maybe you should have posted about how Debian doesn't offer to run updates by defaults but gives you the option to do so if you like.
Oh, and MacOSX and RedHat takes the same attitude (up2date utility sits in corner, turns red if there's an update you should apply, MacOS has a similar tool).::makes a raspberry::
Circa Windows 2000, service pack 3. By default, this already happens.
The story here is that Microsoft backed off when privacy groups thought this was a crummy idea (especially with the EULA of SP3 and XP SP1, big-brother visions abound).
Now they are saying they'd consider giving you more control over this, and to, by default, accept security-relevant patches in this manner by default. Also, (big item), they'll ship the machines with the firewall enabled. That alone is probably the best idea they've adopted under recent community pressure.
Do you remember STREAMS? If you've done any Solaris kernel programming you might have bumped heads with it.
Think like the DirectShow API on Windows and you're getting close. It's a bunch of kernel and user space "modules" that have pins that fit into each other, complete with control pins that can pass certain kinds of data. Pins can send interrupts, etc, req/initiate block transfers, etc.
STREAMS is a lot like that, and it'd make sense to make the video drivers and stuff STREAMS interface instead of standard block/char devices.
I understand there was an initiative to brings STREAMS to linux (LiS) but it so far has not been widely accepted.
If SVR4 streams don't catch on and replace these outdated APIs, I am working on a proposal for such a system that could go beyond what STREAMS would provide... a system that combines the features of demand loaded modules, DirectX, GConf, and system libraries. A generic, run-time configurable multimedia/multiformat API and MPI, with hardware/driver support where sensible.
you can buy a generic 100mpbs card based on rt_8139 or something similar. I've never had a problem with these cards, and they offload a decent amount from the CPU.
I've always had trouble with 3Com. Did you ever see a 3Com 3C9xx try to talk to 3Com stackers? It ain't pretty.
They got Creative to release full exposure on its EMU10k and the Audigy. (I don't know the status of the Audigy 2)
They got fairly exclusive access to ATI's specs. They have Matrox all over the board. NVidia is playing proxy to whatever users want (which is fine).
Intel provides generous documentation, etc.
Look at the Alsa project! They managed to rattle quite a few cages and get support for arguably more sound hardware then Windows XP supports.
In fact, it's the periphial manufacturers who give people a harder time. Getting firewire or serial ATA support is like pulling teeth. Most modems' drivers are reverse engineered from windows drivers and external debugging hardware.
There are companies out there that will build you a fit-perfect Linux box along the lines of an Apple. They'll tell you what hardware works and doesn't. But the PC world is wild and wooly so people will want to roam free and buy unsupported products and try to make them work.
Since OSS is about choice, we shouldn't try to get people to NOT buy things. OSS is not in a position to write most any firmware because the manufacturers often regard such things as trade secret, and sometimes the devices do not meet certain regulations (IE the 802.11a/g drivers). This is unfortunate.
Also, thanks for the recognition. You'd be surprised how easily you get modded up when you coyly represent the elicit fantasy of many a Slashdotter.
But why not this behavior if POSIXLY_CORRECT is not set?
test the first argument (if it's the only one), and if it looks like a tar file, untars it to the current directory (if it's - it reads it from standard input). otherwise, it tars up the path/filespec to standard output.
That sounds like the "most obvious" usage in absence of the "command".
other examples besides tar and ar? dump, restore, mknod. All commands I have to pull up the manpage for (that and mencoder and sox, but I digress)
mousedev.c in the USB HID system tries to coalesce all mouse-type events into the IMPS/2 format so that applications that are used to reading straight PS/2 scroll mouse data from the kernel still work, ala X, or gpm.
This only gives you three buttons and two virtual buttons from the scroll wheel. The side buttons disappear (examine the code, BTN_3 and BTN_4 "fall through" to middle and right click).:-( Either that or force it to run in "explorerps/2" mode.
I wish someone would hack that driver to expose the mouse events in/dev/input/mouseev or something like that, because XFree is PERFECTLY CAPABLE of reading raw usb hid frames from a mouse. So then you could have 10 billion buttons and tehy'd all work.
the ide-scsi thing still happens with cdburning. But most modern distros include that by default and activate it when they detect cd-burning capable hardware (RedHat, SusE, Mandrake, etc.) In fact, there's no reason why not to make all your drives ide-scsi (unless you REALLY didn't want SCSI in the kernel for some reason).
The whole AGP/APIC/NVidia thing mostly stems from a whole shitload of dense hardware specs that are impossible to implement, and partially complete support on AMD chipsets. NVidia tries to push things hard, and then shit hits the fan when the drivers for the PCI/AGP bridges aren't all quite there. At least with Windows, they have decent APIC support and the motherboard manufacturer can fine tune a driver release for it. In linux, most motherboard manufacturers will ignore requests for specs (because it might reveal the shoddiness in how ACPI is implemented, or whatever)
That is often a symptom of having a slightly mentally-challenged AGP controller/northbridge. Your kernel might need to have some option turned on to deal with a CRAAAAZY chipset.
You should be able to specify an the NFS-related "soft" and "intr" options to volumes that are "user" mountable (i.e. CDROM, floppy, and digital cameras)
For the filesystems that support it, it would be neat if they allow you to kill apps holding the disc open, or suspend the filesystem (while making the disc look busy), allowing you to eject the disc (so long as you don't try to access files on it). Later replacement of the disc with another would invalide the entire VFS cache and "magically" update it with a new filesystem underneath. Not sure would be the behavior of the threads with already open files and working directories (hold buffers until it closes? make readdir fail always?)
Really the bug is with the application. If it must be terminated to properly eject the disk, then you shouldn't use it. (this is no exception, MacOSX and Windows NT have these issues as well).
Enough rope to hang yourself with. And the kernel module loader seems to take care of most everything nowadays, so they rarely see any action.
If it isn't automagical, point your complaint at the driver author. (Sometimes it's the 3rd party guys trying to help out! Bad! Bad! Buy the O'Reilly book on writing device drivers first!)
WinZIP is an application suite that handles many compression formats.
GZIP is a single compression format. It can only handle gzipped files (duh!). If it handled more, it wouldn't be a tiny utility, and that wouldn't be very unix-like, would it? GZIP needs to stay small because it's used in tiny places like initial RAM disks and boot floppies.
WinZIP actually uses the library in gzip to handle.gz files, imagine that. WinZIP is BIG.
Search freshmeat for archiving utilities (with names that often sound like linzip or similar). These are what you are really looking for. Also note that later Nautulis (gnome-vfs) and Konqueror release can browse into many types of archives as if they were folders.
arguments to single-letter options occur in the order in which they are specified. Thus, in tar cvf, f requires an argument, which follows the cvf cluster, but is BEFORE the files to tar.
Similarly, if you were to, say, exclude something, you might do this:
tar -cvfX foo.tar./file/to/exclude./files_to_tar
but!
tar -cvXf./file/to/exclude foo.tar./files_to_tar
notice the correlation between the order of arguments, and the options that go with them. The files to process are ALWAYS last.
The following syntax are also valid:
tar -cv -f foo.tar -X./file/to/exclude./files_to_tar tar -fX foo.tar./file/to/exclude -cv./files_to_tar etc.
Note that each option cluster starts with a '-', and any options are slurped in to "complete" them.
This is the standard for all unix commands. Where've you been?
Note: the LEGITIMATE complain about tar that I can understand is that it always assumes the first option is an option cluster even if it doesn't start with '-'. You would think it'd just collect the arguments and tar them to standard input, but you'd be mistaken. That always bothered me. The first file will be treated as a cluster, with often disastrous results. Yea for POSIX compliance
Basis for Small Browsing OS or Car MP3 player?
on
Windows 95 in 4.47MB
·
· Score: 2, Insightful
Has anyone not heard of QNX demo disks, ZipSlack, or the many pre-made Car MP3 linux distros?
People, it's not that hard! Why suffer with a non-pre-emptive, not protected-mode OS to do these common things?
Backwards compatibiltiy, and games, I can understand. A cool Win95 tiny system would be a great way to bootstrap an old DOS or DirectX game on a CD.
But for car MP3 players? I hope you like the music to crap out every so often, if you change tracks too quickly. You'd best be running CubePlayer with some kinda custom input controller if you want it to work predictably.
And I quote: Do you think the freeloader mentality on the Internet is ready for change?
I think it's at the turn of the hockey stick, because it's at about 15 percent of the Web population that's paying for content right now--that's still a low number. Very soon, you'll see that the content that's left to be free is content that will not be trusted; content that has a bias. Just like when you pick up a magazine that's free, and you don't trust it.
If this isn't a reality-distortion-field view of the Internet and web publication I don't know what is.
15%? THAT's highballing it. I bet those numbers were garnered from a website poll that sells subscription services (and sells ads to display at subscribers).
Pretty soon free content will be perceived as biased? If that ever occurs, the Internet is officially dead. Might as well call it CableTV 2.0. Whatever happened to the idea of giving as well as taking? If everyone who cared to voice an honest opinion had a webserver, then everyone would be contributing equally, and paying for Internet access would be payment enough.
Now I'm pissed off. I can't believe bozos like this are still spinning up this dredge. I hope he goes down in a flaming ball of VC cash.
Slashdot Mirror
You're a piggy.
::shimmies and shakes::
Wooooooooooooo!
The focus of the article is that Microsoft is ignoring people who cry "privacy" and opt for updates by default with a choice to disable or screen because they cannot allow Slammer worms to thrive when patches have been out for months.
::makes a raspberry::
Also, they are going to ship with firewalls enabled. People who want to run servers will have to learn to open ports.
Microsoft is adopting a secure-by-default stance, even if it inconviences some users or ruffles a few feathers.
The fact that Debian can do these things has nothing to do with it. I take it you are very proud of Debian, but maybe you should have posted about how Debian doesn't offer to run updates by defaults but gives you the option to do so if you like.
Oh, and MacOSX and RedHat takes the same attitude (up2date utility sits in corner, turns red if there's an update you should apply, MacOS has a similar tool).
Slashdot, I present you the reverse-psychology troll!
Do you troll for people who play the devil's advocate, or think they have a non-conformist point of view or what? I'm really confused.
Circa Windows 2000, service pack 3.
By default, this already happens.
The story here is that Microsoft backed off when privacy groups thought this was a crummy idea (especially with the EULA of SP3 and XP SP1, big-brother visions abound).
Now they are saying they'd consider giving you more control over this, and to, by default, accept security-relevant patches in this manner by default.
Also, (big item), they'll ship the machines with the firewall enabled. That alone is probably the best idea they've adopted under recent community pressure.
GSSE - the Generic Service Supplication Engine.
Do you remember STREAMS? If you've done any Solaris kernel programming you might have bumped heads with it.
Think like the DirectShow API on Windows and you're getting close. It's a bunch of kernel and user space "modules" that have pins that fit into each other, complete with control pins that can pass certain kinds of data. Pins can send interrupts, etc, req/initiate block transfers, etc.
STREAMS is a lot like that, and it'd make sense to make the video drivers and stuff STREAMS interface instead of standard block/char devices.
I understand there was an initiative to brings STREAMS to linux (LiS) but it so far has not been widely accepted.
If SVR4 streams don't catch on and replace these outdated APIs, I am working on a proposal for such a system that could go beyond what STREAMS would provide... a system that combines the features of demand loaded modules, DirectX, GConf, and system libraries. A generic, run-time configurable multimedia/multiformat API and MPI, with hardware/driver support where sensible.
Sigh, if only.
you can buy a generic 100mpbs card based on rt_8139 or something similar. I've never had a problem with these cards, and they offload a decent amount from the CPU.
I've always had trouble with 3Com. Did you ever see a 3Com 3C9xx try to talk to 3Com stackers? It ain't pretty.
You were saying?
They don't use SNMP to manage Crays. I should know, I work for SGI.
YHL, HAND F4GG075-EXP3rT!!1
They got Creative to release full exposure on its EMU10k and the Audigy. (I don't know the status of the Audigy 2)
They got fairly exclusive access to ATI's specs. They have Matrox all over the board. NVidia is playing proxy to whatever users want (which is fine).
Intel provides generous documentation, etc.
Look at the Alsa project! They managed to rattle quite a few cages and get support for arguably more sound hardware then Windows XP supports.
In fact, it's the periphial manufacturers who give people a harder time. Getting firewire or serial ATA support is like pulling teeth. Most modems' drivers are reverse engineered from windows drivers and external debugging hardware.
There are companies out there that will build you a fit-perfect Linux box along the lines of an Apple. They'll tell you what hardware works and doesn't. But the PC world is wild and wooly so people will want to roam free and buy unsupported products and try to make them work.
Since OSS is about choice, we shouldn't try to get people to NOT buy things. OSS is not in a position to write most any firmware because the manufacturers often regard such things as trade secret, and sometimes the devices do not meet certain regulations (IE the 802.11a/g drivers). This is unfortunate.
Also, thanks for the recognition. You'd be surprised how easily you get modded up when you coyly represent the elicit fantasy of many a Slashdotter.
But why not this behavior if POSIXLY_CORRECT is not set?
test the first argument (if it's the only one), and if it looks like a tar file, untars it to the current directory (if it's - it reads it from standard input). otherwise, it tars up the path/filespec to standard output.
That sounds like the "most obvious" usage in absence of the "command".
other examples besides tar and ar? dump, restore, mknod. All commands I have to pull up the manpage for (that and mencoder and sox, but I digress)
And alsa usually fixes it.
You don't have a slipstream install of SP3? That enables DMA by default, and fixes a few file permission fuckups.
but the default settings are saner (Windows tends to "center" the sliders if you haven't set them)
mousedev.c in the USB HID system tries to coalesce all mouse-type events into the IMPS/2 format so that applications that are used to reading straight PS/2 scroll mouse data from the kernel still work, ala X, or gpm.
:-( Either that or force it to run in "explorerps/2" mode.
/dev/input/mouseev or something like that, because XFree is PERFECTLY CAPABLE of reading raw usb hid frames from a mouse. So then you could have 10 billion buttons and tehy'd all work.
This only gives you three buttons and two virtual buttons from the scroll wheel. The side buttons disappear (examine the code, BTN_3 and BTN_4 "fall through" to middle and right click).
I wish someone would hack that driver to expose the mouse events in
the ide-scsi thing still happens with cdburning. But most modern distros include that by default and activate it when they detect cd-burning capable hardware (RedHat, SusE, Mandrake, etc.) In fact, there's no reason why not to make all your drives ide-scsi (unless you REALLY didn't want SCSI in the kernel for some reason).
The whole AGP/APIC/NVidia thing mostly stems from a whole shitload of dense hardware specs that are impossible to implement, and partially complete support on AMD chipsets. NVidia tries to push things hard, and then shit hits the fan when the drivers for the PCI/AGP bridges aren't all quite there. At least with Windows, they have decent APIC support and the motherboard manufacturer can fine tune a driver release for it. In linux, most motherboard manufacturers will ignore requests for specs (because it might reveal the shoddiness in how ACPI is implemented, or whatever)
You spent $2000 on that iBook. There's a reason for that.
Linux users have no such luxury to be able to test every piece of hardware in every system with all its little ideosyncracies.
[n/t]
That is often a symptom of having a slightly mentally-challenged AGP controller/northbridge. Your kernel might need to have some option turned on to deal with a CRAAAAZY chipset.
You should be able to specify an the NFS-related "soft" and "intr" options to volumes that are "user" mountable (i.e. CDROM, floppy, and digital cameras)
For the filesystems that support it, it would be neat if they allow you to kill apps holding the disc open, or suspend the filesystem (while making the disc look busy), allowing you to eject the disc (so long as you don't try to access files on it). Later replacement of the disc with another would invalide the entire VFS cache and "magically" update it with a new filesystem underneath. Not sure would be the behavior of the threads with already open files and working directories (hold buffers until it closes? make readdir fail always?)
Really the bug is with the application. If it must be terminated to properly eject the disk, then you shouldn't use it. (this is no exception, MacOSX and Windows NT have these issues as well).
Enough rope to hang yourself with. And the kernel module loader seems to take care of most everything nowadays, so they rarely see any action.
If it isn't automagical, point your complaint at the driver author. (Sometimes it's the 3rd party guys trying to help out! Bad! Bad! Buy the O'Reilly book on writing device drivers first!)
WinZIP is an application suite that handles many compression formats.
.gz files, imagine that. WinZIP is BIG.
GZIP is a single compression format. It can only handle gzipped files (duh!). If it handled more, it wouldn't be a tiny utility, and that wouldn't be very unix-like, would it? GZIP needs to stay small because it's used in tiny places like initial RAM disks and boot floppies.
WinZIP actually uses the library in gzip to handle
Search freshmeat for archiving utilities (with names that often sound like linzip or similar). These are what you are really looking for. Also note that later Nautulis (gnome-vfs) and Konqueror release can browse into many types of archives as if they were folders.
arguments to single-letter options occur in the order in which they are specified. Thus, in tar cvf, f requires an argument, which follows the cvf cluster, but is BEFORE the files to tar.
Similarly, if you were to, say, exclude something, you might do this:
tar -cvfX foo.tar
but!
tar -cvXf
notice the correlation between the order of arguments, and the options that go with them. The files to process are ALWAYS last.
The following syntax are also valid:
tar -cv -f foo.tar -X
tar -fX foo.tar
etc.
Note that each option cluster starts with a '-', and any options are slurped in to "complete" them.
This is the standard for all unix commands. Where've you been?
Note: the LEGITIMATE complain about tar that I can understand is that it always assumes the first option is an option cluster even if it doesn't start with '-'. You would think it'd just collect the arguments and tar them to standard input, but you'd be mistaken. That always bothered me. The first file will be treated as a cluster, with often disastrous results. Yea for POSIX compliance
Has anyone not heard of QNX demo disks, ZipSlack, or the many pre-made Car MP3 linux distros?
People, it's not that hard! Why suffer with a non-pre-emptive, not protected-mode OS to do these common things?
Backwards compatibiltiy, and games, I can understand. A cool Win95 tiny system would be a great way to bootstrap an old DOS or DirectX game on a CD.
But for car MP3 players? I hope you like the music to crap out every so often, if you change tracks too quickly. You'd best be running CubePlayer with some kinda custom input controller if you want it to work predictably.
And I quote:
Do you think the freeloader mentality on the Internet is ready for change?
I think it's at the turn of the hockey stick, because it's at about 15 percent of the Web population that's paying for content right now--that's still a low number. Very soon, you'll see that the content that's left to be free is content that will not be trusted; content that has a bias. Just like when you pick up a magazine that's free, and you don't trust it.
If this isn't a reality-distortion-field view of the Internet and web publication I don't know what is.
15%? THAT's highballing it. I bet those numbers were garnered from a website poll that sells subscription services (and sells ads to display at subscribers).
Pretty soon free content will be perceived as biased? If that ever occurs, the Internet is officially dead. Might as well call it CableTV 2.0.
Whatever happened to the idea of giving as well as taking? If everyone who cared to voice an honest opinion had a webserver, then everyone would be contributing equally, and paying for Internet access would be payment enough.
Now I'm pissed off. I can't believe bozos like this are still spinning up this dredge. I hope he goes down in a flaming ball of VC cash.
??? Is that supposed to be Satan then?