Re:Looking forward... mostly
on
Quicksilver
·
· Score: 0
oh poop! you mean cromagnon man remains aren't hobbit bones?
Re:Looking forward... mostly
on
Quicksilver
·
· Score: 1
haha, I believe that's a parody of some of Star Trek fan mail....like "it took 10 seconds in episode 25 for the transporter to dematerialize a person, but only 7 seconds in episode 27 What, have they improved the transporter?"
heh, I wasn't complaining, I love GNU/Linux and it's amazing to watch it develop at 3x the rate of commercial Unix with some of the smartest designers and coders on earth. The open source solutions for all those things are int the works. Right now I have to do commercial + open source mix, but I know the free replacements are in the works. Heck, in two years I know Oracle will get its lunch eaten by Postgresql...the core of Oracle is such an obsolete, archaic, stick-shift piss poor VSAM era design mush that I can't wait to stuff that crap in the crapper where it belongs.
Some proprietary products require a certain size of swap as a certified configuration
yup, Linux has an LVM, but I've had it puke on me before in test lab; we'd have never used it in production. I know LInux will soon have all the data center features I listed plus more. Two more years, I think.
The swap limit is with 2.4 kernel on Intel 32 bit processors [it's hard-coded in memory/paging ]. Other architectures might go higher - if you are on 32 bit intel with 2.4 kernel I think you might not really have what the outputs to free indicate
maybe Sun was hoping some scientific/engineering application would try it. I only meant Sun has demonstrated it. No business is going to run such a weird beast, as any additional CPU over 72 has to go on I/O channels. I couldn't find anyone who has bought one. Fujitsu does make a 128-way Sparc PrimePower server "done right", but I never heard of any U.S. customers.
Sadly, biggest Sun I've actually touched and worked with is 24-way Sunfire 6800.
Good grief, who is going to buy the weirdo Solaris for x86? All the software for sparc isn't available on x86 side. It's slow. Device drivers for major devices are missing; you'll need to be careful to run it on THEIR approved configurations. It's expensive compared to $0 for Linux (excluding support, which you don't get a meaningful level of that with just base OS purchase anyway). Don't get me wrong, I like Solaris: on a Sparc where it belongs , for things Linux can't do yet . Give Linux two more years, and all the high end datacenter features will be done & perfected.
Sun says 128 CPUs in their PDF document "Datasheet: the Solaris 9 operating system".
You can read about 2.4 and 2.6 SMP scalability here Though Linux can run on 64-way, it is currently best on 8-way or less, with 16 and 32-way improvements still in the works
Both FreeBSD and Linux started SMP with very coarse mutex methods because it's very HARD to write that stuff. They will get better over time. In Linux, IBM is helping to tune and improve that stuff (and SCO hates it and wants to claim it)
heh, I should have also added I've been paid to do Oracle PSQL and LISP and COBOL/Java integration in the last 10 years.......which Visual Studio product would I use to do that?
I can't answer how I know about MS source code control problems. But it shows in their product problems
Our mothers could run NeXTSTEP or Mac System 6,7 or OS/X just fine.
Visual studio?????????? How would that help me write platform independent code on operating systems that I have been PAID to write for: Solaris, IRIX, Linux, AIX, HP/UX as well as Windows? I can assure you far superior development environments exist for any langauge in common use you care to name (C, C++, Objective C, Java), except perhaps Visual Basic (which is of no interest/use/monetary gain to me)
I too have spent over 10+ years administering SunOS/Solaris and 5 years with RedHat's distribution of GNU/Linux. I would just add for rock solid stability on the *low* end approaching that of Solaris one should probably use FreeBSD or OpenBSD, not Linux.
What is Linux as of today (2.4.x kernel, 2.6 isn't ready yet!) missing for higher end servers?
Hot plugging for SCSI devices that is reliable (adding and removing can be a mixed bag, it does't always work for all types of devices, especially in SAN situation)
Reliable open source volume manager that is rock solid
distributed lock management
size of single swap partition limited to 2GB
high performance filesystem that is also solid. All the journalled filesystems available on Linux can have inconsistency after crash at just the wrong time; also, too many journalling threads can bring system to its knees as during Oracle load. Let's just get a good FFS for Linux already!
People LIKE ME pay alot of money for microsoft software. And for my hard earned money, and my company's money, I get insecure crap that is flawed by stupid design (like for instance, executing attachements AS A FEATURE).
All my open source software cost me $0, and has never once let an intruder in or brought my system to its knees.
We have a right to bash. I've wasted weeks of my life because of Microsoft's poor quality control and lack of source code control.
The things you are talking about could exist on top of Linux, any other Unix, OS/400, or just about any other modern operating system. It's not that Linux isn't ready for the desktop, the problem is at a higher level. The GUI isn't Linux, just as the GUI you happen to run on OS X isn't OS X (you can install other ones).
I will agree with you that Mac leads in the GUI area (NeXTStep was even better for awhile); I remember being able to walk up to a macintosh in 1985 and use it with no instruction whatever (and the current OSX is much better than what existed then) Still, some training is still needed for most people who aren't "computer people"
Pachyderm just denotes a "thick-skinned mammal"...it's really not a good biological classification at all. If you check your dictionary, hippos and rhinos and elephants are included, and some dictionaries even show the word used for hogs!
I know very well about that; it proves my point. Especially the bit at the end:
Faced with this reality, the IEEE Standards Board decided that IEEE standards will use the conventional, internationally adopted, definitions of the SI prefixes. Mega will mean 1 000 000, except that the base-two definition may be used (if such usage is explicitly pointed out on a case-by-case basis) until such time that prefixes for binary multiples are adopted by an appropriate standards body.
actually, it's we computer people who invented the very FALSE notion that 120 gig is 120 x 1024 x 1024 x 1024. We redefined (being lazy S.O.B.'s) the meaning of gig from its true meaning of 10^9 to 2^30, mega from 10^6 to 2^20, kilo from 10^3 to 2^10. So now marketers find it to their advantage to use the TRUE meaning of the word and you cry "lies"? The majority opinion of science and engineering votes against us.
Really, the only problem is that there is no standard to do this with our existing filesystems. We already have the capability to do this a myriad of ways with ANY filesystem in common use
actually, he's doing hot fusion - but by his neutron output it's clear that not even a a couple dozen deuterium atoms per minute are fusing. So the energy distribution is such that ALMOST NO ATOMS in his device have the needed energy to fuse.
Architects are licensed by the states here in the U.S., and much building code knowledge is quite region specific. Lawyers and engineers too are licensed by state, with regional knowledge also a factor. This makes it more difficult to push these professions overseas than most IT jobs.
I would think any IT certifications or licensing would allow global participation, on the other hand.
he had an analog information storage-replication-retrieval system of audio information, which might be called a "computer" in the broad sense of machinery to transform information.
Slashdot Mirror
oh poop! you mean cromagnon man remains aren't hobbit bones?
haha, I believe that's a parody of some of Star Trek fan mail....like "it took 10 seconds in episode 25 for the transporter to dematerialize a person, but only 7 seconds in episode 27 What, have they improved the transporter?"
heh, I wasn't complaining, I love GNU/Linux and it's amazing to watch it develop at 3x the rate of commercial Unix with some of the smartest designers and coders on earth. The open source solutions for all those things are int the works. Right now I have to do commercial + open source mix, but I know the free replacements are in the works. Heck, in two years I know Oracle will get its lunch eaten by Postgresql...the core of Oracle is such an obsolete, archaic, stick-shift piss poor VSAM era design mush that I can't wait to stuff that crap in the crapper where it belongs.
Some proprietary products require a certain size of swap as a certified configuration
yup, Linux has an LVM, but I've had it puke on me before in test lab; we'd have never used it in production. I know LInux will soon have all the data center features I listed plus more. Two more years, I think.
The swap limit is with 2.4 kernel on Intel 32 bit processors [it's hard-coded in memory/paging ]. Other architectures might go higher - if you are on 32 bit intel with 2.4 kernel I think you might not really have what the outputs to free indicate
maybe Sun was hoping some scientific/engineering application would try it. I only meant Sun has demonstrated it. No business is going to run such a weird beast, as any additional CPU over 72 has to go on I/O channels. I couldn't find anyone who has bought one. Fujitsu does make a 128-way Sparc PrimePower server "done right", but I never heard of any U.S. customers.
Sadly, biggest Sun I've actually touched and worked with is 24-way Sunfire 6800.
Good grief, who is going to buy the weirdo Solaris for x86? All the software for sparc isn't available on x86 side. It's slow. Device drivers for major devices are missing; you'll need to be careful to run it on THEIR approved configurations. It's expensive compared to $0 for Linux (excluding support, which you don't get a meaningful level of that with just base OS purchase anyway). Don't get me wrong, I like Solaris: on a Sparc where it belongs , for things Linux can't do yet . Give Linux two more years, and all the high end datacenter features will be done & perfected.
in theory? The 15K is sold with up to 106 processors. Sun claims 128 CPU's max for "in theory" in their literature (see my other reply)
Sun says 128 CPUs in their PDF document "Datasheet: the Solaris 9 operating system".
You can read about 2.4 and 2.6 SMP scalability here Though Linux can run on 64-way, it is currently best on 8-way or less, with 16 and 32-way improvements still in the works
Both FreeBSD and Linux started SMP with very coarse mutex methods because it's very HARD to write that stuff. They will get better over time. In Linux, IBM is helping to tune and improve that stuff (and SCO hates it and wants to claim it)
heh, I should have also added I've been paid to do Oracle PSQL and LISP and COBOL/Java integration in the last 10 years.......which Visual Studio product would I use to do that?
I can't answer how I know about MS source code control problems. But it shows in their product problems
Our mothers could run NeXTSTEP or Mac System 6,7 or OS/X just fine.
Visual studio?????????? How would that help me write platform independent code on operating systems that I have been PAID to write for: Solaris, IRIX, Linux, AIX, HP/UX as well as Windows? I can assure you far superior development environments exist for any langauge in common use you care to name (C, C++, Objective C, Java), except perhaps Visual Basic (which is of no interest/use/monetary gain to me)
What is Linux as of today (2.4.x kernel, 2.6 isn't ready yet!) missing for higher end servers?
People LIKE ME pay alot of money for microsoft software. And for my hard earned money, and my company's money, I get insecure crap that is flawed by stupid design (like for instance, executing attachements AS A FEATURE).
All my open source software cost me $0, and has never once let an intruder in or brought my system to its knees.
We have a right to bash. I've wasted weeks of my life because of Microsoft's poor quality control and lack of source code control.
The things you are talking about could exist on top of Linux, any other Unix, OS/400, or just about any other modern operating system. It's not that Linux isn't ready for the desktop, the problem is at a higher level. The GUI isn't Linux, just as the GUI you happen to run on OS X isn't OS X (you can install other ones).
I will agree with you that Mac leads in the GUI area (NeXTStep was even better for awhile); I remember being able to walk up to a macintosh in 1985 and use it with no instruction whatever (and the current OSX is much better than what existed then) Still, some training is still needed for most people who aren't "computer people"
Pachyderm just denotes a "thick-skinned mammal"...it's really not a good biological classification at all. If you check your dictionary, hippos and rhinos and elephants are included, and some dictionaries even show the word used for hogs!
yes, it's great being #1. We do whatever the hell we want, and the rest of the world can only complain!
I know very well about that; it proves my point. Especially the bit at the end:
Faced with this reality, the IEEE Standards Board decided that IEEE standards will use the conventional, internationally adopted, definitions of the SI prefixes. Mega will mean 1 000 000, except that the base-two definition may be used (if such usage is explicitly pointed out on a case-by-case basis) until such time that prefixes for binary multiples are adopted by an appropriate standards body.
See, we should call the 2^30 type giga an "Imperial giga", while the rest of science and engineering can use the wee little 10^9 "standard giga".
actually, it's we computer people who invented the very FALSE notion that 120 gig is 120 x 1024 x 1024 x 1024. We redefined (being lazy S.O.B.'s) the meaning of gig from its true meaning of 10^9 to 2^30, mega from 10^6 to 2^20, kilo from 10^3 to 2^10. So now marketers find it to their advantage to use the TRUE meaning of the word and you cry "lies"? The majority opinion of science and engineering votes against us.
the coconut technically isn't a nut, it's a dry nupe.
Really, the only problem is that there is no standard to do this with our existing filesystems. We already have the capability to do this a myriad of ways with ANY filesystem in common use
actually, he's doing hot fusion - but by his neutron output it's clear that not even a a couple dozen deuterium atoms per minute are fusing. So the energy distribution is such that ALMOST NO ATOMS in his device have the needed energy to fuse.
Verisign doesn't have a history of abiding by the law, if you remember their domain expiration scam.
Architects are licensed by the states here in the U.S., and much building code knowledge is quite region specific. Lawyers and engineers too are licensed by state, with regional knowledge also a factor. This makes it more difficult to push these professions overseas than most IT jobs.
I would think any IT certifications or licensing would allow global participation, on the other hand.
or a sign of someone who knows latin (his use is proper)
he had an analog information storage-replication-retrieval system of audio information, which might be called a "computer" in the broad sense of machinery to transform information.