Running tests after installing a new kernel sounds like a good idea, but it would be better if they could be automated. Hardware tests might be better run by hand, since they risk damaging cheap chop suey hardware, but for software tests (like Python, Mesa etc. mentioned on that page) it'd be possible to have the tests installed and ready to run.
What I mean is that every package in the distribution installs its test suite in/usr/test/ or somewhere, so after doing an upgrade you can just 'test everything' with a single command. As well as kernels, that would be handy for checking that a library upgrade hasn't broken anything which uses that library, a new shell hasn't broken any shell scripts, and so on.
Re:Servers were never allowed out on cable
on
Broadband Crackdown
·
· Score: 1
If each user gets his own IP address (whether a real address or some 192.168.x.y masqueraded thing), then it's possible to limit the number of packets or total traffic flow to/from each address. Yes, the physical cable itself is shared, like Ethernet, but someone has to route the packets to the Internet and at that place you can do whatever quotaing you wish.
Re:Servers were never allowed out on cable
on
Broadband Crackdown
·
· Score: 3, Insightful
If the bandwidth is limited, then quota the bandwidth to each user! It's just as possible to eat up the limited upstream bandwidth by uploading large files to Hotmail, but they don't ban that.
OK, how many of the so-called Users on this site are actually pseudonyms for Malda and his cronies? Perhaps I am one myself.
Re:KDE 2s2 feature depth is astounding
on
KDE 2.2 Tagged
·
· Score: 1
Closing a window is kinda destructive. Maybe it's not such a bad thing that you have to deliberately position the mouse over the X icon before clicking. Just be thankful that KDE doesn't make you type 'y e s ' a la Emacs:-).
Well yes, the patent licence is about usage of the so-called 'invention', rather than about distributing copies. That's how patents work. But since the only condition is that the patent be used as part of a GPL (or similar) program, and any version of Linux fits this criterion, no extra restrictions are imposed *in practice* by the patent licence. Anything which used some code from Linux would have to be GPLed anyway.
That sounds okay... the patent is licensed freely for use in any GPLed software. This would appear to conform to the GPL's provisions about patents (basically, you may not distribute the program without also granting a licence for any applicable patents) and it looks reasonable from common sense.
It's not demanding any special fee for commercial use, so it counts as free software still (aka Open Source etc etc).
At least, from the paragraph you quoted above everything seems fine. It would still be better for everyone if patent offices would refrain from granting monopolies on abstract ideas however...
Wait.. You think the lower left corner is great for Ctrl, since you can hold it down easily, but it's an awkward place
for Meta?
You have a point. I started trying to type Emacs stuff on my XT keyboard, putting my thumb on the Alt key, since that's what I'd do on an ordinary PC keyboard (without Windows keys). Of course putting your thumb on Alt is a bad idea if it's over in the corner, you should use your little finger instead. But I didn't think that far, I just thought 'hey, this is really awkward' and gave up. Duuh.
I'd pay big bucks for an old IBM AT keyboard in good condition.. I still have my IBM PC XT's keyboard even
though it unfortunately won't work with modern computers (though the very next generation IBM made, the
aforementioned AT keyboard, is very similar and does work with modern machines)
I believe that although the connector (5-pin DIN) looks similar, the electronics are completely different. It's a real pity that nobody makes XT-to-AT convertors (or if they did once get made, that they're no longer available).
You know these things:
good solid click when you type
Well lots of keyboards will give you a solid click, for example the various IBM PS/2 keyboards. which are what I use most of the time. The difference with the XT is the _metallic_, ringing echo you get after hitting each key... it sounds almost like a futuristic typewriter:-).
the ` key next to the ' key
Strange... I have a couple of XT keyboards here which have no backtick character at all. Oh, I know why, they're the UK version (livre sign on Shift-3). BTW the ` and ' are not symmetric characters in modern character sets, so it's arguably a misfeature to make them look like they are (unless you use TeX a lot). I don't know whether the PC-XT had them as left and right quote characters or as backtick and upright quote.
the \ key on the left, mirroring the / key
My PS/2 keyboards and clone keyboards still have this. Again this must be due to the British keyboard layout, which I believe is a variant on the general 'International' PS/2 layout. (At first I was annoyed that adding one currency symbol to the keyboard should cause several other things to move about randomly, but there are some redeeming features like the bigger Enter key. I wonder whether the 'US International' keyboard layout has these things. The Acorn Archimedes keyboard is an example of how to include a pound sign without screwing about with the location of other keys.)
the * key near the., so it was easy to type stuff like "*.txt"
Again, my XT keyboard doesn't have this; * is on Shift-8 just as on modern keyboards.
Ctrl on the left, Caps Lock way the fuck out of the way where you could never accidentally hit it
Actually, for Emacs use I think I probably prefer Ctrl in the corner, so you can hold it down with your little finger and tap out X C, X S, or whatever. Holding down the key to the left of A feels more awkward.
Esc next to 1
Of course what's really needed is *zero* next to one...
A gigantic spacebar that goes all the way from Alt (which is where Left Ctrl is on modern keyboards)
Again, this is a bit awkward for use as a Meta key in Emacs. But I agree, the big spacebar is cool.
If you're trusting the network without doing any proper checks, that's your problem. Somebody could plug in his own PC and start spoofing IP packets _today_. The release of WinXP doesn't change that.
What about the 'only root can use ports 1024' feature of Unixes, which Windows doesn't implement? Does that mean that Windows is a security threat? No. If you're being so stupid as to trust the originating port number, you deserve everything you get.
Egress and, er, ingress filtering around the edge of your network may be good enough most of the time; it doesn't protect you against PCs inside the network starting to spoof things, but you may feel you can trust your own employees (and don't let them run Outlook).
I assume you mean 'megaFLOPS hour', in other words million floating point operations per second, for one hour. That's a silly unit (as is the 'kilowatt hour' used for electricity). Rather than dividing by seconds and then multiplying by hours, it would be better to measure per gigaFLOP (3.6 gigaFLOP = 1 megaFLOPS hour).
This reminds me of copy protection schemes for floppy disks that worked by deliberate corruption. Changing the checksum for a particular sector of the disk, or something, so it would appear that any read had failed. It wasn't done at the filesystem level because even a 'raw backup' would fail.
I remember thinking at the time, I wish this machine would stop trying to be helpful and check the validity of what it's reading, and instead just give me the data with no questions asked.
I know that CDs use some kind of Gray code or other ECC to encode 16-bit sample values into 20-bit words or something similar. Then there are other error-correction measures, checksums and so on. That's why a CD holds only 650Mbyte (or a bit more) although the physical capacity in terms of raw bits is much higher.
Is there any software or hardware to give a genuinely 'raw' CD image, before any of the error correction has been performed? Such an image would probably be around a gigabyte in size.
I think I would buy three of these things, export the disks as network block devices, and then do software RAID-5 on them from a fourth machine. Or something like that...
I thought people on Slashdot wanted Bruce Campbell (Autolycus in Xena and Hercules) to replace Mulder. That didn't happen, so they picked Lawless as a second choice?
I wrote a small utility that lets you say: amount cp/mnt/floppy/*.txt somewhere. It mounts whatever's needed before running the command, and unmounts it afterwards.
It irks me that I have to log in with a password and then give another password for my ssh keypair. Someone started writing a PAM module that would let you log in if you could give the correct passphrase for your ssh key; have any Linux distributions started using it?
I've said before that file sharing programs need to move to https as soon as possible. That's one port ISPs couldn't block, if they want their customers to be able to buy stuff online, and because it's secure nobody can inspect the traffic to find out whether it's file sharing or just web browsing (except with unreliable heuristics on amount of data transferred, etc).
There will be set operations on characters classes: for instance, you'll be able to specify a match of all letters without the vowels. Actually, this will probably be in Perl 5 as well. There was also time to tell us that he'd like the/x modifier to be on by default.
I'd like to see such operators applied to regexps as a whole, not just character classes. We already have the | operator - the regular expression a|b matches anything that a matches _or_ anything that b matches.
But how about a\-b for something which fits a but not b? I've often wished for that when writing a complex regexp, it would make life so much simpler. Also a\&b for something matched by both a _and_ b would be very handy.
I think you may already be able to do some of this stuff with Perl's zero-width lookahead assertions, but it's a bit messy and not particularly obvious.
Is this definitely _the_ book to have on writing device drivers? I know there are other books written about the Linux kernel, but it seems that this is currently the best choice if you want to create a device driver.
Anyone disagree - are there any alternative books covering this topic?
Is the Kylix development environment itself GPLed? If not, you'd be using a proprietary editor/compiler/whatever, but you'd be obliged to release your own code under the GPL (unless you pay for the commercial edition). Kinda seems like the worst of both worlds.
Yes, I was trolling. My point is that there's shedloads of stuff which 'costs nothing' to install, and some of it is actually pretty good. There's no reason why Java should get special treatment. OTOH, ship a half-decent perl5 environment and people could really get some useful work done:-).
We should also demand that manufacturers ship a copy of Cygwin with each Windows PC. It costs them nothing but makes it much easier to download and run lots of free software ported to Windows.
Heck, while you're at it, why not 'demand' that they include a Linux partition? Hard disk space is cheap these days.
Slashdot Mirror
Running tests after installing a new kernel sounds like a good idea, but it would be better if they could be automated. Hardware tests might be better run by hand, since they risk damaging cheap chop suey hardware, but for software tests (like Python, Mesa etc. mentioned on that page) it'd be possible to have the tests installed and ready to run.
/usr/test/ or somewhere, so after doing an upgrade you can just 'test everything' with a single command. As well as kernels, that would be handy for checking that a library upgrade hasn't broken anything which uses that library, a new shell hasn't broken any shell scripts, and so on.
What I mean is that every package in the distribution installs its test suite in
If each user gets his own IP address (whether a real address or some 192.168.x.y masqueraded thing), then it's possible to limit the number of packets or total traffic flow to/from each address. Yes, the physical cable itself is shared, like Ethernet, but someone has to route the packets to the Internet and at that place you can do whatever quotaing you wish.
If the bandwidth is limited, then quota the bandwidth to each user! It's just as possible to eat up the limited upstream bandwidth by uploading large files to Hotmail, but they don't ban that.
OK, how many of the so-called Users on this site are actually pseudonyms for Malda and his cronies? Perhaps I am one myself.
Closing a window is kinda destructive. Maybe it's not such a bad thing that you have to deliberately position the mouse over the X icon before clicking. Just be thankful that KDE doesn't make you type 'y e s ' a la Emacs :-).
Well yes, the patent licence is about usage of the so-called 'invention', rather than about distributing copies. That's how patents work. But since the only condition is that the patent be used as part of a GPL (or similar) program, and any version of Linux fits this criterion, no extra restrictions are imposed *in practice* by the patent licence. Anything which used some code from Linux would have to be GPLed anyway.
That sounds okay... the patent is licensed freely for use in any GPLed software. This would appear to conform to the GPL's provisions about patents (basically, you may not distribute the program without also granting a licence for any applicable patents) and it looks reasonable from common sense.
It's not demanding any special fee for commercial use, so it counts as free software still (aka Open Source etc etc).
At least, from the paragraph you quoted above everything seems fine. It would still be better for everyone if patent offices would refrain from granting monopolies on abstract ideas however...
You have a point. I started trying to type Emacs stuff on my XT keyboard, putting my thumb on the Alt key, since that's what I'd do on an ordinary PC keyboard (without Windows keys). Of course putting your thumb on Alt is a bad idea if it's over in the corner, you should use your little finger instead. But I didn't think that far, I just thought 'hey, this is really awkward' and gave up. Duuh.
If you're trusting the network without doing any proper checks, that's your problem. Somebody could plug in his own PC and start spoofing IP packets _today_. The release of WinXP doesn't change that.
What about the 'only root can use ports 1024' feature of Unixes, which Windows doesn't implement? Does that mean that Windows is a security threat? No. If you're being so stupid as to trust the originating port number, you deserve everything you get.
Egress and, er, ingress filtering around the edge of your network may be good enough most of the time; it doesn't protect you against PCs inside the network starting to spoof things, but you may feel you can trust your own employees (and don't let them run Outlook).
I assume you mean 'megaFLOPS hour', in other words million floating point operations per second, for one hour. That's a silly unit (as is the 'kilowatt hour' used for electricity). Rather than dividing by seconds and then multiplying by hours, it would be better to measure per gigaFLOP (3.6 gigaFLOP = 1 megaFLOPS hour).
This reminds me of copy protection schemes for floppy disks that worked by deliberate corruption. Changing the checksum for a particular sector of the disk, or something, so it would appear that any read had failed. It wasn't done at the filesystem level because even a 'raw backup' would fail.
I remember thinking at the time, I wish this machine would stop trying to be helpful and check the validity of what it's reading, and instead just give me the data with no questions asked.
I know that CDs use some kind of Gray code or other ECC to encode 16-bit sample values into 20-bit words or something similar. Then there are other error-correction measures, checksums and so on. That's why a CD holds only 650Mbyte (or a bit more) although the physical capacity in terms of raw bits is much higher.
Is there any software or hardware to give a genuinely 'raw' CD image, before any of the error correction has been performed? Such an image would probably be around a gigabyte in size.
I think I would buy three of these things, export the disks as network block devices, and then do software RAID-5 on them from a fourth machine. Or something like that...
I thought people on Slashdot wanted Bruce Campbell (Autolycus in Xena and Hercules) to replace Mulder. That didn't happen, so they picked Lawless as a second choice?
I wrote a small utility that lets you say: amount cp /mnt/floppy/*.txt somewhere. It mounts whatever's needed before running the command, and unmounts it afterwards.
So you mean that with Debian, you just have to enter one password (your ssh passphrase) and it logs you in based on that alone?
Or perhaps you just mean that Debian automatically starts ssh-agent for you on login (which other distributions, eg Mandrake, do as well).
Not using a null passphrase is exactly what the article at IBM warns you about!
It irks me that I have to log in with a password and then give another password for my ssh keypair. Someone started writing a PAM module that would let you log in if you could give the correct passphrase for your ssh key; have any Linux distributions started using it?
Many (if not most) https sites don't have a certificate.
I've said before that file sharing programs need to move to https as soon as possible. That's one port ISPs couldn't block, if they want their customers to be able to buy stuff online, and because it's secure nobody can inspect the traffic to find out whether it's file sharing or just web browsing (except with unreliable heuristics on amount of data transferred, etc).
I'd like to see such operators applied to regexps as a whole, not just character classes. We already have the | operator - the regular expression a|b matches anything that a matches _or_ anything that b matches.
But how about a\-b for something which fits a but not b? I've often wished for that when writing a complex regexp, it would make life so much simpler. Also a\&b for something matched by both a _and_ b would be very handy.
I think you may already be able to do some of this stuff with Perl's zero-width lookahead assertions, but it's a bit messy and not particularly obvious.
Is this definitely _the_ book to have on writing device drivers? I know there are other books written about the Linux kernel, but it seems that this is currently the best choice if you want to create a device driver.
Anyone disagree - are there any alternative books covering this topic?
Is the Kylix development environment itself GPLed? If not, you'd be using a proprietary editor/compiler/whatever, but you'd be obliged to release your own code under the GPL (unless you pay for the commercial edition). Kinda seems like the worst of both worlds.
Yes, I was trolling. My point is that there's shedloads of stuff which 'costs nothing' to install, and some of it is actually pretty good. There's no reason why Java should get special treatment. OTOH, ship a half-decent perl5 environment and people could really get some useful work done :-).
We should also demand that manufacturers ship a copy of Cygwin with each Windows PC. It costs them nothing but makes it much easier to download and run lots of free software ported to Windows.
Heck, while you're at it, why not 'demand' that they include a Linux partition? Hard disk space is cheap these days.