I don't know about the pipe lock, but for containers this guy at Sun put 190 containers on a 31GB disk. It is possible to cut them down, at least.
The man page for SMF says it's database is 'transactional' and is able to provide old configurations. I haven't done this, but it seems troubleshooting is possible. Also, there are a bunch of XML files in/var for SMF that are human readable. SMF is just a big step from the traditional init mechanism, and it will take a while for people to get used to it.
Recently, I've started getting the itch to upgrade to 1 Gig in my home computer (up from 512MB). This is just for normal day-to-day use with several apps open at once. 4 Gig home PCs are really only a few years away, I think.
"It looks like they hired too many windows programers to deal with core items in the OS and that means security features I've been using for years can be walked around."
Security is easier under Solaris 10. I was able to enable TCP wrappers with one SMF command line, for example. The SMF command line is really a work-saver in the long term, but the rc.d directories are still there for compatibility.
"A brand new v100 can't cope with drives bigger than 128 gig under solaris."
A v100 is not brand new. It's a years-old design. Get a v210 or a v20z.
"The old exploits for the named pipes still can trash a system since deadlocking pid 1 is always good thing."
The only pipe I see under/etc is accessible only by root. How is that exploitable without some other exploit, first?
"The new services thing is windows registry meets init and cron and rc."
Not really. The Windows Registry is a terrible hack job that causes Windows users tons of grief. Rc scripts are just tedious and slow. Cron is still there as a separate utility. Sun's SMF is pretty simple once you get a hang of it, and it has the ability to re-start services automatically if they crash. I really don't see the parallel you claim at all.
"The new containers require so many packages that its just wrong."
Containers are optional, and Sun introduced a new smaller install option for network servers.
"However the my risk analysis say the disadvantages outweigh the advantages so there will be no more new sun orders from my company."
My analysis is that you were seeking excuses to fit your established bias.
Actually, I think the Linux desktops really are getting there. I'm running JDS3 (on Solaris, but it's basically still GNOME/OO.org), and I really need Windows for very very few things. Given that the major Linux desktops are also much cheaper than Windows for similar capability, I'd say Microsoft is really worried but not showing it.
Just like AMD pulled a fast one on Intel, Linux/UNIX will do the same to Microsoft. My hope is that "Wintel" will be a paragraph in the history books and little more.
Duke was probably planning this switch long before Solaris 10 and OpenSolaris rolled around. Solaris 10 is what UNIX should have been years ago, if only because of DTrace (DTrace is really really cool, and would be great for a senior-level elective).
Some companies have said that if Sun was doing three years ago what they are doing now (Solaris 10, OpenSolaris, free licensing), they would not have switched to Linux. Consider that Sun still guarantees binary and source compatibility when migrating to Solaris 10 from older versions, while Linux cannot. Linux is very useful, but there are still things that make long-term deployments awkward at times. Mod what you will, but it is true.
1) "Fire" employees from Company A. 2) Said employees set up Company B that de-compiles competitor's product and writes a specification for it. 3) Company A writes new software based on the specification. 4) The Company B dissolves. 5) Employees are re-hired to Company A after the management had a "change of heart."
No, they all do it. CNN, FOX, MSNBC, CBS, ABC, NBC. They all focus on drama rather than news, and they all have political biases of some form. The only TV news I can stand watching any more is on PBS, where they take the biases and stick them in a moderated debate without trying to make it "infotainment."
Please explain how Linus writing Linux was not reverse-engineering of Unix.
Well, aren't all the interfaces published? There's POSIX, IEEE, etc. Further, the Linux kernel mimics UNIX but doesn't re-implement everything (last I checked a Linux module doesn't just plug into FreeBSD or Solaris).
I'd use the words "inspired by UNIX" rather than "reverse engineered from UNIX."
Can you copyright the *idea* of the algorithm? The theif could re-write everything to avoid traditional word-for-word copyright, and I'm not sure the original researcher could do much once the cat's out of the bag. At a minimum, the original researcher would have to get a good lawyer and deal with a huge interruption in their life. The alternative would be to get a patent, but patents aren't appropriate for all situations. This is where good security and a tight lip are the best policy.
"'IP' is specious word with no real bearing or precise meaning."
I generally consider IP to be trade secrets, copyright, patents, and the public domain. If someone isn't set up properly in the former, they irrevocably live in the latter. "Theft", in a sense, is moving things into the public domain without permission.
Another non-corporate example: imagine being a researcher at a university. You develop a radical new algorithm that takes a O(n^3) process and make it into O(n log n). This algorithm is of great importance in, say, fluid dynamics or something really time-consuming. Unfortunately, you are prepping your work for publication and due credit, when someone breaks in and steals your files and publishes under a different name first. Since you have not published, yet, there really is no protection at all, and you just lost two years of work.
1. you said "IP" suggesting that it is a tangible thing that can be stolen
If IP is a trade secret, than "stealing" means that what was once private is now public. The real theft, then, is a denial of value to the company regarding that IP. Just because it isn't something tangible like jewelry doesn't make it less of a crime.
I don't see how this would protect them, as copyright protection doesn't imply protection of trade secrets, which is what the submitter is probably concerned about. The only real protection for trade secrets is trusting employees, and an NDA might be appropriate in the employment contract. The key isn't to remove all of the technology from the offices, but to create enough dis-incentives to prevent the employees from wanting to steal.
dual core chips are just that - two cpu's in one packaging
IMO, the smarter way to view dual core is just as a continuation of moores law. Instead of doubling the performance of a single core, they put a second core into the same overall space along with the same overall power consumption. It's about time software vendors suck it up and charge per socket.
Not really. The Opterons are: faster and cooler than Intel's chips.
There was a news leak this week about Sun shipping an 8-way PCI-Express-based Opteron server later this year. With dual-core, that's basically a 16-way server with a shitload of bandwidth--in 4U.
A 16-way server of Xeons is kind-of a joke, right now.
Look part way down for "CDDL is file-based; that means that files licensed under the CDDL can be combined with files licensed under other licenses, whether open source or proprietary."
"The GPL doesn't favour any project or any developer."
The GPL favors GPL projects and GPL developers.
If those projects are a mix of CDDL, MPL, et at, we can't do that.
Please show me how code under the BSD license, CDDL, and MPL are impossible to combine in a single project (e.g,, one linked into a single executable binary). My understanding is that they can be combined with no problems, given licenses don't mix within individual files.
"Suppose I want to take some Solaris code and put it in Darwin."
You create a module that contains the Solaris code and link it into the final program. What in the licenses prohibits this?
Slashdot Mirror
I don't know about the pipe lock, but for containers this guy at Sun put 190 containers on a 31GB disk. It is possible to cut them down, at least.
The man page for SMF says it's database is 'transactional' and is able to provide old configurations. I haven't done this, but it seems troubleshooting is possible. Also, there are a bunch of XML files in
Recently, I've started getting the itch to upgrade to 1 Gig in my home computer (up from 512MB). This is just for normal day-to-day use with several apps open at once. 4 Gig home PCs are really only a few years away, I think.
"It looks like they hired too many windows programers to deal with core items in the OS and that means security features I've been using for years can be walked around."
/etc is accessible only by root. How is that exploitable without some other exploit, first?
Security is easier under Solaris 10. I was able to enable TCP wrappers with one SMF command line, for example. The SMF command line is really a work-saver in the long term, but the rc.d directories are still there for compatibility.
"A brand new v100 can't cope with drives bigger than 128 gig under solaris."
A v100 is not brand new. It's a years-old design. Get a v210 or a v20z.
"The old exploits for the named pipes still can trash a system since deadlocking pid 1 is always good thing."
The only pipe I see under
"The new services thing is windows registry meets init and cron and rc."
Not really. The Windows Registry is a terrible hack job that causes Windows users tons of grief. Rc scripts are just tedious and slow. Cron is still there as a separate utility. Sun's SMF is pretty simple once you get a hang of it, and it has the ability to re-start services automatically if they crash. I really don't see the parallel you claim at all.
"The new containers require so many packages that its just wrong."
Containers are optional, and Sun introduced a new smaller install option for network servers.
"However the my risk analysis say the disadvantages outweigh the advantages so there will be no more new sun orders from my company."
My analysis is that you were seeking excuses to fit your established bias.
Users who need to write a simple HTTP server, for example, can't use SunRays because they'll collide.
Okay, SunRay + Solaris Containers?
Sun's website says that SunRay can run on Linux, too, so the GP post might be correct.
The other part of me realizes that there's no need for expensive Sun hardware for public terminals...
SunRay?
Actually, I think the Linux desktops really are getting there. I'm running JDS3 (on Solaris, but it's basically still GNOME/OO.org), and I really need Windows for very very few things. Given that the major Linux desktops are also much cheaper than Windows for similar capability, I'd say Microsoft is really worried but not showing it.
Just like AMD pulled a fast one on Intel, Linux/UNIX will do the same to Microsoft. My hope is that "Wintel" will be a paragraph in the history books and little more.
Duke was probably planning this switch long before Solaris 10 and OpenSolaris rolled around. Solaris 10 is what UNIX should have been years ago, if only because of DTrace (DTrace is really really cool, and would be great for a senior-level elective).
Some companies have said that if Sun was doing three years ago what they are doing now (Solaris 10, OpenSolaris, free licensing), they would not have switched to Linux. Consider that Sun still guarantees binary and source compatibility when migrating to Solaris 10 from older versions, while Linux cannot. Linux is very useful, but there are still things that make long-term deployments awkward at times. Mod what you will, but it is true.
There was an e-mail circulating a while back with a photo of a NASCAR fan who shaved Dale Earnhardt's number into his back hair. Irresistable...
What about this:
1) "Fire" employees from Company A.
2) Said employees set up Company B that de-compiles competitor's product and writes a specification for it.
3) Company A writes new software based on the specification.
4) The Company B dissolves.
5) Employees are re-hired to Company A after the management had a "change of heart."
No, they all do it. CNN, FOX, MSNBC, CBS, ABC, NBC. They all focus on drama rather than news, and they all have political biases of some form. The only TV news I can stand watching any more is on PBS, where they take the biases and stick them in a moderated debate without trying to make it "infotainment."
Please explain how Linus writing Linux was not reverse-engineering of Unix.
Well, aren't all the interfaces published? There's POSIX, IEEE, etc. Further, the Linux kernel mimics UNIX but doesn't re-implement everything (last I checked a Linux module doesn't just plug into FreeBSD or Solaris).
I'd use the words "inspired by UNIX" rather than "reverse engineered from UNIX."
But copyright is on your side.
Can you copyright the *idea* of the algorithm? The theif could re-write everything to avoid traditional word-for-word copyright, and I'm not sure the original researcher could do much once the cat's out of the bag. At a minimum, the original researcher would have to get a good lawyer and deal with a huge interruption in their life. The alternative would be to get a patent, but patents aren't appropriate for all situations.
This is where good security and a tight lip are the best policy.
"'IP' is specious word with no real bearing or precise meaning."
I generally consider IP to be trade secrets, copyright, patents, and the public domain. If someone isn't set up properly in the former, they irrevocably live in the latter. "Theft", in a sense, is moving things into the public domain without permission.
Another non-corporate example: imagine being a researcher at a university. You develop a radical new algorithm that takes a O(n^3) process and make it into O(n log n). This algorithm is of great importance in, say, fluid dynamics or something really time-consuming. Unfortunately, you are prepping your work for publication and due credit, when someone breaks in and steals your files and publishes under a different name first. Since you have not published, yet, there really is no protection at all, and you just lost two years of work.
"And they actually produce work?"
You ask this in a post to Slashdot...amusing.
1. you said "IP" suggesting that it is a tangible thing that can be stolen
If IP is a trade secret, than "stealing" means that what was once private is now public. The real theft, then, is a denial of value to the company regarding that IP. Just because it isn't something tangible like jewelry doesn't make it less of a crime.
I don't see how this would protect them, as copyright protection doesn't imply protection of trade secrets, which is what the submitter is probably concerned about. The only real protection for trade secrets is trusting employees, and an NDA might be appropriate in the employment contract. The key isn't to remove all of the technology from the offices, but to create enough dis-incentives to prevent the employees from wanting to steal.
Wow, moderated into the gutter. What's the agenda, here?
dual core chips are just that - two cpu's in one packaging
IMO, the smarter way to view dual core is just as a continuation of moores law. Instead of doubling the performance of a single core, they put a second core into the same overall space along with the same overall power consumption. It's about time software vendors suck it up and charge per socket.
Sounds like some VERY hefty competition for AMD.
Not really. The Opterons are: faster and cooler than Intel's chips.
There was a news leak this week about Sun shipping an 8-way PCI-Express-based Opteron server later this year. With dual-core, that's basically a 16-way server with a shitload of bandwidth--in 4U.
A 16-way server of Xeons is kind-of a joke, right now.
Also, IANAL, but the CDDL itself frequently talks about "larger works", where a "larger work" can contain non-CDDL code.
OpenSolaris FAQ
Look part way down for "CDDL is file-based; that means that files licensed under the CDDL can be combined with files licensed under other licenses, whether open source or proprietary."
"The GPL doesn't favour any project or any developer."
The GPL favors GPL projects and GPL developers.
If those projects are a mix of CDDL, MPL, et at, we can't do that.
Please show me how code under the BSD license, CDDL, and MPL are impossible to combine in a single project (e.g,, one linked into a single executable binary). My understanding is that they can be combined with no problems, given licenses don't mix within individual files.
"Suppose I want to take some Solaris code and put it in Darwin."
You create a module that contains the Solaris code and link it into the final program. What in the licenses prohibits this?