My father used to sell Electrolux vacuum cleaners when I was very young, and he had an older coworker who gave him this sage advice on romantic relations with people at work: never deficate where you eat.
Now, we are not talking romantic relations in the above subject, but even so relations with friends will have their ups and downs, and it is unfair to an employer to expose their work environment to the emotional instabilities in your or your friend's lives.
I have had close personal friends at work in the past, but these days I keep a healthy distance, and things go much easier. There are lots of people who are suitable friends out there; your workplace is not your only option. YMMV.
If we mandated the use of fire-proof building materials, incredible amounts of damage to personal property might be avoided, but the costs of retrofitting the building industry are excessive.
But... if you absolutely, positively, cannot tolerate a fire, then you probably want a "VMS" house (if you don't mind writing your shell scripts in FORTRAN).
In the same way, object filesystems, system-wide GC, system-wide bounds checking, and permissions that support a more complex model than ugo would lead to fewer system cracks.
The war will go on for years but IBM has secured quite a beachhead in the opening salvo.
Most owners of a PC can tell you whether they run an Intel, AMD, VIA, Transmeta or other CPU. The same is not true for game consoles.
Console makers have used many processors, among them Z80s, 68000, MIPS, Intel, et al. Users remained blissfully ignorant of the architecture of their systems.
IBM is conquering a market where brand loyalty is not possible. If Intel elects to enter this market and pours r&d money towards the effort, can IBM hold them off? It remains to be seen if this will be a Phyrric victory.
Use VMS if you like shell scripting in FORTRAN
on
DECnet Isn't Dead
·
· Score: 1, Interesting
While Dave Cutler (perpitrator of a great many OS atrocities) once remarked that "UNIX is a junk OS designed by a committee of PH.D.s," his operating systems have some profound problems.
Can anyone argue that VMS DCL has evolved as much as the Bourne environment? I believe it was Dennis Ritchie who severely criticised VMS for integrating most of the command interpreter into the kernel (which Cutler again did by moving many drivers from Ring 3 to Ring 0 in NT - same mistake?).
Yes, VMS has awesome capability in clustering and security, far outpacing most past and future implementations, but much of it has evolved about as far as JCL in becoming a modern system.
Lots of things just simply can't be done on VMS. UNIX is much more of a "happy medium" and has proven to be highly adaptable.
...as it quickly degenerates. Capitalism seems to me to be an intermediate state between anarchy and oligarchy.
Just as the distribution of particles in a chemical suspension will not remain uniform without constant agitation, the distribution of economic power will not remain disbursed throughout the population without constant antitrust enforcement.
It seems that we are seeing the tail end of this now.
The GNU project also produces gcc, which is used by all of the free *nixes to compile their code.
Yes, but we are talking about Linux kernel code quality, not gcc code quality, which is another discussion entirely. Questioning Linux is not the same as questioning GNU.
Perhaps we should talk about the history of ssh a bit here... As for using an `alternate, less tested sshd', are you sure you don't work for Microsoft's FUD department or something?
So, off the top of your head, how many other sshd implementations implement privilege separation? How many commercial UNIX distributions bundle an alternate sshd?
The OpenBSD project also produces OpenSSH, and is kind enough to package portable versions that run on alternative architectures.
This is a thankless task for Theo & company; I think that the largest corporate dontation for OpenSSH was one laptop from IBM. Shameful.
Theo can run his systems free of all code produced by Linus. Unless Linus wants to go back to telnet (or use an alternate, less tested sshd), he has no such option.
If you are connecting your system to the internet in any way, OpenBSD's virtues shine. Propolice and w^x are features that are rarely implemented in mainstream Linux distributions, I positively cannot live without spamd mail greylisting, the patch traffic is easily a tenth of RedHat if not less, the utilities are better (a real inetd, a real ksh), and the security otherwise just sings. On the downside, there is no lvm, no pam, some of the packages are older (samba) and some features aren't available.
That said, there are plenty of places where I will deploy CentOS - if I need an Oracle database, if I'm running a complex Linux application or I need to compile Linux binaries, or if I might need to eventually be on a formally supported platform (RedHat Enterprise).
There are other times when I might choose something else (NetBSD because it supports some wierdo hardware, SuSE because an application is certified to it, Mac OS X because it arouses my curiousity, etc.). I would hesitate in moving many of my systems this way, though.
Sometimes the practical concerns can drive the choice of platform pretty hard - moreso than most might think.
...and even the founding fathers had trouble with it. Alexander Hamilton asserted that a republic was a state with no special privilege due to birth or family. James Madison felt uncomfortable with this definition (as would any rational southerner) as he owned slaves. The classical republic of antiquity, the republic of Rome, even appointed dictators from time to time.
This word "republic" is fluid, and it mostly means (IMHO) "land without a king." Most people who use it (including members of the US Republican party) invoke the word without any idea of the ambiguity that it implies.
I don't like the Patriot Act either, but we aren't to the point where we have to fear being killed for critizing our leadership or laws either.
Who did Kennedy criticise? Why did we find Ron Brown, chairman of the Democratic National Committee, in the wreckage of a plane with a bullet in his head?
This is the US, and suspicious death is our middle name.
OpenBSD had GCC 2.9 and 3.x for a while. I think 3.7 is the first release in a while with just one C compiler.
Each platform had only one compiler, and the kernel was built with that same compiler. Red Hat 6 came with two different compilers on each platform.
What, like the IPF to PF transition?
I started using OpenBSD at 3.2, and this was already done, so it was no pain to me. Implementing PF was a sound choice, it caused a major version upgrade (2 to 3), it removed a big licensing problem, and it was obviously the correct way to handle the situation. I would much rather have the IPF to PF transition than the ipfwadm to ipchains to iptables mess.
Or the a.out to ELF transition that broke all the old binaries?
I knew that this was coming, and I planned for it. Yes, a major version upgrade was probably called for, but it didn't really change anything from an administration perspective.
There is less reason to review OpenBSD.
on
OpenBSD 3.7 Reviewed
·
· Score: 3, Insightful
When I ran RedHat, there were some pretty annoying things that got changed from release to release (inetd disappears, two different C compiler installs because of kernel problems, etc.).
This kind of stuff doesn't happen in OpenBSD. From an administration perspective, my first 3.2 install is very similar to the 3.5 that I run now, which itself is similar to 3.7. There are no large architecture changes (perhaps because things are well-thought-out from the start).
Because of this, you pretty much know what you're getting when a new OpenBSD release comes out. The installer is practically identical, and the running system yeilds few surprises. There will always be new features, but there won't be lots of things to unlearn.
So no, I don't really pay much attention to the reviews. The list of new features on the OpenBSD web page pretty much tells me all that I need to know.
While many devices are not supported, and the performance is not good, HURD/Mach is feature complete (and most of Debian runs on it, IIRC).
Because the performance was bad, the new HURD effort focuses on reimplementing on L4. Perhaps with a faster microkernel, Apple could have avoided the kludge of an in-kernel BSD peer.
If I am reading correctly, Mach is responsible for IPC in the Apple kernel. It would be interesting to see benchmarks of SYSV system calls to semaphores, queues, and shared memory (and perhaps even basic signal handling) under Linux and the Apple kernel on the same hardware if those are entirely handled by Mach.
However, I've based my own statements on comments from the KHTML team.
Apple's individual diffs are not visible; Apple's source management tool will most likely be required to see these.
KHTML members have commented on (the difficulty of removing) OSX-specific code; I don't mean to offend, but echo these statements.
Perhaps several branches will be required for a fully-integrated KHTML-Safari tree - STABLE, Apple-production, and development. One way or another, one side will constantly be folding in changes produced by the other. Perhaps they should all switch to Bitkeeper?
An Apple high-level executive should be arranging meetings with KHTML core to sort this out - the longer they delay, the worse it will appear.
I can imagine that the Gecko core is breathing a collective sigh of relief that they were not chosen. Perhaps the ability to fork with impunity influenced Apple's choice of KHTML.
I'm not so sure that Webcore is better than KHTML
on
Safari vs. KHTML
·
· Score: 1
While the gobstopping patches are a problem, specific mention has been made of code quality concerns in moving patches back into KHTML.
Apple is deadline-oriented, and this will sometimes imply sloppy code.
...then it's still useless to me. I'll keep my crufty old pdksh.
Apple should start...
on
Safari vs. KHTML
·
· Score: 3, Insightful
...by getting everybody on the same source control software.
AFAIK, KHTML uses CVS, and Apple internally uses Perforce.
Nothing constructive can be done until everything is on the same platform.
Apple, offer to buy licenses of your source control software for the KHTML core. Even if they still spurn you, it will appear to the rest of us that you at least tried. You will look more and more of a villan until you make some effort at a reconciliation.
p.s. The KHTML team will need to be conversant with OSX to the point that they can remove GUI calls to it and replace them with QT. If this is a current problem, then some books might be in order.
...is that there is a line in the new testament that says that Jesus will come again "before those who are now living have died." The Rapture, being slightly short of 2000 years late, is then called into question.
However, perhaps this was only referring to the Beloved Disciple, who Jesus referred to at the end of the book of John by saying "what is it to you if he lives until I come again?"
In any case, the texts of the New Testament are obviously heavily edited. They are drawn from other sources (the Gospel of Thomas and the lost "Q" Gospel), in some areas they bear the mark of the council of Nicea, and IMHO the broad message is more important than any specific detail.
Is receiving the number of the best an unforgivable sin? Jesus said in Matthew that all sin could be forgiven except the sin of "blaspheming the Holy Ghost" - however, this text itself may be a Nicea edit in reaction to the Arian Heresy. Are those who died damned lost forever? If so, then why did Peter say that Jesus preached to the dead in the time between the crucifixion and the resurrection?
The book makes you think for yourself. It's amazing how many Christians have never bothered to acquaint themselves with the details and incongruities. I can hardly claim great knowledge, but I can make a lot of heads spin with the little that I know.
Slashdot Mirror
There are definite consequences to sexual relations at work. Some are written down, some are not.
My father used to sell Electrolux vacuum cleaners when I was very young, and he had an older coworker who gave him this sage advice on romantic relations with people at work: never deficate where you eat.
Now, we are not talking romantic relations in the above subject, but even so relations with friends will have their ups and downs, and it is unfair to an employer to expose their work environment to the emotional instabilities in your or your friend's lives.
I have had close personal friends at work in the past, but these days I keep a healthy distance, and things go much easier. There are lots of people who are suitable friends out there; your workplace is not your only option. YMMV.
The Russians can already get 3 people into space reliably; if I have read this correctly, then what they lack is the heavy lifter.
Should NASA concentrate its efforts on this heavy lifter and leave crew transport to the Russians for the moment?
Reducing costs through coordination seems reasonable to me.
...and so does UNIX, from time to time.
If we mandated the use of fire-proof building materials, incredible amounts of damage to personal property might be avoided, but the costs of retrofitting the building industry are excessive.
But... if you absolutely, positively, cannot tolerate a fire, then you probably want a "VMS" house (if you don't mind writing your shell scripts in FORTRAN).
In the same way, object filesystems, system-wide GC, system-wide bounds checking, and permissions that support a more complex model than ugo would lead to fewer system cracks.
We don't do it because we're lazy.
cli and gui continue to make many mistakes. The OS to fix these first wins my vote (even if it's M$):
In general, gui and cli have ignored one another for too long. Share some ideas!
Most owners of a PC can tell you whether they run an Intel, AMD, VIA, Transmeta or other CPU. The same is not true for game consoles.
Console makers have used many processors, among them Z80s, 68000, MIPS, Intel, et al. Users remained blissfully ignorant of the architecture of their systems.
IBM is conquering a market where brand loyalty is not possible. If Intel elects to enter this market and pours r&d money towards the effort, can IBM hold them off? It remains to be seen if this will be a Phyrric victory.
While Dave Cutler (perpitrator of a great many OS atrocities) once remarked that "UNIX is a junk OS designed by a committee of PH.D.s," his operating systems have some profound problems.
Can anyone argue that VMS DCL has evolved as much as the Bourne environment? I believe it was Dennis Ritchie who severely criticised VMS for integrating most of the command interpreter into the kernel (which Cutler again did by moving many drivers from Ring 3 to Ring 0 in NT - same mistake?).
Yes, VMS has awesome capability in clustering and security, far outpacing most past and future implementations, but much of it has evolved about as far as JCL in becoming a modern system.
Lots of things just simply can't be done on VMS. UNIX is much more of a "happy medium" and has proven to be highly adaptable.
...as it quickly degenerates. Capitalism seems to me to be an intermediate state between anarchy and oligarchy.
Just as the distribution of particles in a chemical suspension will not remain uniform without constant agitation, the distribution of economic power will not remain disbursed throughout the population without constant antitrust enforcement.
It seems that we are seeing the tail end of this now.
Yes, but we are talking about Linux kernel code quality, not gcc code quality, which is another discussion entirely. Questioning Linux is not the same as questioning GNU.
...that Linux owes to Theo.
The OpenBSD project also produces OpenSSH, and is kind enough to package portable versions that run on alternative architectures.
This is a thankless task for Theo & company; I think that the largest corporate dontation for OpenSSH was one laptop from IBM. Shameful.
Theo can run his systems free of all code produced by Linus. Unless Linus wants to go back to telnet (or use an alternate, less tested sshd), he has no such option.
If you are connecting your system to the internet in any way, OpenBSD's virtues shine. Propolice and w^x are features that are rarely implemented in mainstream Linux distributions, I positively cannot live without spamd mail greylisting, the patch traffic is easily a tenth of RedHat if not less, the utilities are better (a real inetd, a real ksh), and the security otherwise just sings. On the downside, there is no lvm, no pam, some of the packages are older (samba) and some features aren't available.
That said, there are plenty of places where I will deploy CentOS - if I need an Oracle database, if I'm running a complex Linux application or I need to compile Linux binaries, or if I might need to eventually be on a formally supported platform (RedHat Enterprise).
There are other times when I might choose something else (NetBSD because it supports some wierdo hardware, SuSE because an application is certified to it, Mac OS X because it arouses my curiousity, etc.). I would hesitate in moving many of my systems this way, though.
Sometimes the practical concerns can drive the choice of platform pretty hard - moreso than most might think.
...and even the founding fathers had trouble with it. Alexander Hamilton asserted that a republic was a state with no special privilege due to birth or family. James Madison felt uncomfortable with this definition (as would any rational southerner) as he owned slaves. The classical republic of antiquity, the republic of Rome, even appointed dictators from time to time.
This word "republic" is fluid, and it mostly means (IMHO) "land without a king." Most people who use it (including members of the US Republican party) invoke the word without any idea of the ambiguity that it implies.
Who did Kennedy criticise? Why did we find Ron Brown, chairman of the Democratic National Committee, in the wreckage of a plane with a bullet in his head?
This is the US, and suspicious death is our middle name.
Microsoft will be supporting Win2k for a total of 10 years.
Red Hat's best support offer is 5 years.
I run OpenBSD. Every release of that gets only a single year.
So how do you plan to convince entrenched M$ shops to switch? This should be interesting.
Each platform had only one compiler, and the kernel was built with that same compiler. Red Hat 6 came with two different compilers on each platform.
I started using OpenBSD at 3.2, and this was already done, so it was no pain to me. Implementing PF was a sound choice, it caused a major version upgrade (2 to 3), it removed a big licensing problem, and it was obviously the correct way to handle the situation. I would much rather have the IPF to PF transition than the ipfwadm to ipchains to iptables mess.
I knew that this was coming, and I planned for it. Yes, a major version upgrade was probably called for, but it didn't really change anything from an administration perspective.
When I ran RedHat, there were some pretty annoying things that got changed from release to release (inetd disappears, two different C compiler installs because of kernel problems, etc.).
This kind of stuff doesn't happen in OpenBSD. From an administration perspective, my first 3.2 install is very similar to the 3.5 that I run now, which itself is similar to 3.7. There are no large architecture changes (perhaps because things are well-thought-out from the start).
Because of this, you pretty much know what you're getting when a new OpenBSD release comes out. The installer is practically identical, and the running system yeilds few surprises. There will always be new features, but there won't be lots of things to unlearn.
So no, I don't really pay much attention to the reviews. The list of new features on the OpenBSD web page pretty much tells me all that I need to know.
While many devices are not supported, and the performance is not good, HURD/Mach is feature complete (and most of Debian runs on it, IIRC).
Because the performance was bad, the new HURD effort focuses on reimplementing on L4. Perhaps with a faster microkernel, Apple could have avoided the kludge of an in-kernel BSD peer.
If I am reading correctly, Mach is responsible for IPC in the Apple kernel. It would be interesting to see benchmarks of SYSV system calls to semaphores, queues, and shared memory (and perhaps even basic signal handling) under Linux and the Apple kernel on the same hardware if those are entirely handled by Mach.
HURD abandoned Mach because of performance issues and is being reimplemented on L4.
If Apple had chosen L4, would it have been necessary from a performance perspective to include BSD at a peer level with the microkernel?
Is it now far too late for Apple to dump Mach?
However, I've based my own statements on comments from the KHTML team.
Perhaps several branches will be required for a fully-integrated KHTML-Safari tree - STABLE, Apple-production, and development. One way or another, one side will constantly be folding in changes produced by the other. Perhaps they should all switch to Bitkeeper?
An Apple high-level executive should be arranging meetings with KHTML core to sort this out - the longer they delay, the worse it will appear.
I can imagine that the Gecko core is breathing a collective sigh of relief that they were not chosen. Perhaps the ability to fork with impunity influenced Apple's choice of KHTML.
While the gobstopping patches are a problem, specific mention has been made of code quality concerns in moving patches back into KHTML.
Apple is deadline-oriented, and this will sometimes imply sloppy code.
...then it's still useless to me. I'll keep my crufty old pdksh.
...by getting everybody on the same source control software.
AFAIK, KHTML uses CVS, and Apple internally uses Perforce.
Nothing constructive can be done until everything is on the same platform.
Apple, offer to buy licenses of your source control software for the KHTML core. Even if they still spurn you, it will appear to the rest of us that you at least tried. You will look more and more of a villan until you make some effort at a reconciliation.
p.s. The KHTML team will need to be conversant with OSX to the point that they can remove GUI calls to it and replace them with QT. If this is a current problem, then some books might be in order.
"Fascism should more properly be called corporatism, since it is the merger of state and corporate power."
...is that there is a line in the new testament that says that Jesus will come again "before those who are now living have died." The Rapture, being slightly short of 2000 years late, is then called into question.
However, perhaps this was only referring to the Beloved Disciple, who Jesus referred to at the end of the book of John by saying "what is it to you if he lives until I come again?"
In any case, the texts of the New Testament are obviously heavily edited. They are drawn from other sources (the Gospel of Thomas and the lost "Q" Gospel), in some areas they bear the mark of the council of Nicea, and IMHO the broad message is more important than any specific detail.
Is receiving the number of the best an unforgivable sin? Jesus said in Matthew that all sin could be forgiven except the sin of "blaspheming the Holy Ghost" - however, this text itself may be a Nicea edit in reaction to the Arian Heresy. Are those who died damned lost forever? If so, then why did Peter say that Jesus preached to the dead in the time between the crucifixion and the resurrection?
The book makes you think for yourself. It's amazing how many Christians have never bothered to acquaint themselves with the details and incongruities. I can hardly claim great knowledge, but I can make a lot of heads spin with the little that I know.
A P120 is overkill for a lot of applications (except recompiling a kernel).