If you choose the "Select Individual Packages" option when installing you can pare it down pretty well. Especially if you don't let it install for failed dependancies.
I've only ever done a default install once, just to see what it was like. It didn't find enough room on the machine.
If you pick packages, you really need to have an idea of what's in them, for instance what -devel packages have which header files is a biggie. The help function to actually describe the package is significantly better than it used to be, but knowing what's in them is still a bit of trial and error.
After you're done, you can also look in the RPM database and deinstall anything extra that you don't like. I've never had difficulty fitting a reasonable X-less install on any old 486 machines I've found.
You're crying wolf. Linux and the associated softwre surrounding it are continually evolving. The kernel release schedule should make that patently obvious. RedHat is keeping as close to up as possible without losing some stability.
Yes, they've made mistakes in the past, so what? You're free to create or use another distribution, fact is that RH's model is working for RedHat, why snivel about it?
Any company large enough to support AIX environments will have at least one person with enough of a clue to (a) wait for stable releases, and (b) FTP. RedHat's done a great deal over the last year to legitimize what we used to have to _sneak_ onto machines. The value in that alone is worth the rest of it to me in my daily job.
Having just gotten capital approval to replace some AIX boxen with both Linux and *BSD machines, and having played the patch/update game for quite a while in both environments (Linux since a.98p11 SLS distribution, AIX for about the same ammount of time) I'm perfectly happy to upgrade my servers when they need it, no matter *which* OS or vendor I happen to use.
If you've paid for AIX and AIX support, you'd know there's no comparison in cost.
I actually tend to buy more CDs personally than we do at work because the release cycle isn't long enough to match my work maintenance cycle. But then I buy more FreeBSD CDs too. FTP installs work just fine from work too, multiple DS-3's make it satisfactory to do.
RedHat also makes updates available for the last couple of releases. Nobody's "forced" to upgrade for bugfixes if they're running a relatively new version-- even if they don't understand that they can apply the patches themselves and compile their own packages.
If you want to complain about people being uninformed, educate them, if you're just whining because RH is being successful, you're more than welcome to try to build your own brand.
RedHat isn't Linux. Linux isn't RedHat. Time you opened *your* eyes to that fact.
> Just look at OpenBSD. It's open source, yet it's > the most secure *nix OS on earth
While OpenBSD is well-done, and fairly secure, it's had exploits available in the past, and probably will have them in the future. I'd go as far as "Damn good security", but nowhere near "most secure." This isn't meant to malign the OBSD team, they've done a significant ammount of very good work.
Sun's Trusted Solaris and HP's Compartmented Mode Workstation (CMW) have a higher level of assurance. I'm not sure where Trusted Irix stands in the stack of things, but it's probably close to or on-par also.
I'd trust Data General's DG/UX- B2 feature set a *lot* more than I'd trust OpenBSD (indeed, NSA seems to be under the impression that it's good enough to house up to SECRET data and still be connected to the Internet- though it's FER seems overdue.)
Finally, I'd also probably put more stock in a properly administered version of Linux with RSBAC (http://www.compuniverse.com/rsbac/) than in OpenBSD because as good as dilligence and auditability are, role enforcement by a security process built into the kernel wins every time.
RSBAC (Ruleset Based Access Control) is an open source project that deserves scrutiny if you're interested in actual security. It's a work in progress that attempts to add B1-type features to the Linux kernel.
> What I'd really like to find is a way to run my > employer's software token in a *Nix environment > so that I don't have to use Windows to dial up
(a) ask the vendor about a Linux port - you never know what you might find out. My Security Dynamics salesdweeb just sent me a Linux executable that's either their client or server code - haven't had time to untar it and look at it yet. But asking for "Whatever you've got that runs on Linux" seems to have paid off with a few hundred K of gzipped tarball.
> to the office. Regrettably, they use a hardware > firewall device whose software clients only come > in one flavor. Short of buying another computer > and using my windows box for a proxy server, I > have no way of connecting a more pleasant OS to > the network where I work
Run NT or 98 in a VMWare window - if all you're using is a software challenge/response (Soft token) program, the VM shouldn't need to connect.
VMWare is $99 for home use, which is cheaper than a new machine. You can also run multiple VMs, so it's good for upgrade testing, testing across versions, etc.
If it has to connect, and you get to do PPP negotiation first, then just IP Masq the VM.
If it's running on something like DOCKMASTER II, it really doesn't need patching. DG/UX B2 is a network-level B2 (under eval, not done) that allows you to define untrusted paths coupled with the IP stack. So, even if you find an exploit, it's likely that the system is set up to not let you reach the level required to execute anything, and the httpd shouldn't be given things like access to a shell. Compartmented systems work fairly well for things like Web servers, nameservers and stuff like that. Data coming in from the Internet is always best set to the lowest level of trust. Web servers, etc. run at the path/user's MAC level/Capability set.
If you're remotely interested in the types of things you can do with OS-level security, do a search for RSBAC to see the types of princilples involved applied to a modern Linux kernel.
I saw something on the Park Police looking at or using cameras that mapped blood vessels in the face a while back. It's my guess that they're using something similar- that'd be much more accurate than a fingerprint scan and significantly harder to fake.
What about making ACs log in as AC and accept a session cookie to post? Valid expired AC session cookies could re-enable later AC access with a cookie exchange, you'd still have anonymous posting, but you'd also have a mechanism that would link multiple posts to a single AC in most instances. Anyone who was moderated down into slashdork status could be redirected to a "behave yourself" page if they still presented the same cookie. Multiple logins from the same IP as an AC within a timeframe could be slowed down like password timeouts to catch scripters - threshold settable per IP even.
> Comon, I dare you to name a few Linux-only > inventions of significant value.
The way Linux does syscalls, which was eventually added to Solaris due to the performance gains would be a good one. Running on systems from 8-bit to 64-bit processors seems to be a good thing that only NetBSD shares with any force. If it wasn't the first 64-bit clean OS, it was certainly one of them. Meanwhile, NT on my Alpha at work still runs in 32-bit mode. Fully-functional packet screening protection in a common kernel seems to have come out of the Open Source community (though perhaps not Linux specificly.)
> Everything in the Linux kernel is a > ripoff.
Bullshit. Most of the code was written from scratch. If you want to talk about processes and the kernel model, then nobody's made significant advances since kernel threading. For that matter, perhaps you can explain what benifits you expect to derrive from this "innovation" in NT? I see a bloated, semi-functional monolithic OS that's so slow that large parts of the user interface had to be moved into kernel space to make it usable on 300Mhz processors. Innovative that was! Maybe you like the innovation of a screen saver taking up 70% of a server's CPU?
> The point Metcalfe is trying to make is, other > companies, like Be, NeXT, Taligent, Sun, Apple, > even Microsoft, are atleast trying to do > something different.
Most of whom are (a) failing, and (b) scrambling to continue the revenue generating mechanisms that keep them afloat. MS does a mediocre job in the desktop OS arena, where it really excells compared to everywhere else it's at. Apple does better, but doesn't have market share. Betamax syndrome.
> Linux is a very simplistic, naive, conservative > implementation of Unix. It's a > monolithic kernel, it doesn't have the concept > of a Yellow box like Mac OS > X, it doesn't have asynchronous or streaming > drivers like Win2k, it doesn't > have a radical filesystem like Be, it doesn't > have a great threading architecture > like Solaris or Be, it's not distributed like > Plan 9. It doesn't have formal > verification, it doesn't have a virtual machine.
And yet every day it does its job just fine. Funnily enough Be doesn't seem to be moving into mission-critical positions. Funnily enough, Beowulf-class clusters seem to have rocketed Linux up the Top 500 supercomputer list without any significant OS-level distribution.
Try writing a driver for Linux then try to write one for NT. Even if you've never written a driver before, Linux will take a few days. Even if you've written drivers before, NT will take weeks.
> Basically, ZERO Innovation, written by a bunch > of part-time hobbyists > copying 30 year old designs with a few stolen > ideas like sendfile()
Really interesting that if we took your belief (wrong though it is) that a bunch of part-time hobbyists seem to be challenging the halls of innovation and winning. Sun, IBM and SGI are all working on Linux support for their OS'. Microsoft is scrambling to try to protect and grow their server market while constantly losing ground to Linux. Good thing it's not a bunch of full time hobbyists if you hold MSFT stock, eh?
> Let's look at what commercial vendors have been > doing:
> o parallel pipeline window systems (years > before X/Precision Insight)
VMs were done well before Next and Taligent were ever companies. Perhaps you forget VM/CMS? As for O-O, it's slow and bloated. Smalltalk proved that case decades ago.
> o microkernel (NT, NeXT, Be)
NT is not a microkernel. The MkLinux port has a real microkernel. It's slow and inefficient compared to a monolithic kernel, hence the wishes of Mac-based Linux users to port the "normal" kernel.
> o fully threaded OS (Be)
pthreads work quite well for user-space programs, but kernel threads are happening.
> o component architecture (NT/COM)
COM sucks, HTH.
> o distributed components built in (DCOM/MTS)
DCOM sucks more, HTH.
> o directory service (Novell, NT)
Um, "So?"
> o plug and play architecture (Win95/98,and > Win2k)
Linux has some plug-and-play features. NT doesn't. Generally it's not as robust or functional as MS would have wished for.
> o relational file system (Be)
Once again, "So?" Journaling file systems have a great deal of value.
> o postscript based WYSIWYG display system > (Display Postscript on the NeXT and Sun NeWS)
This is not an OS feature. Lyx works well as a WYSIWYG thing if you're into page layout though.
> The list just goes on. I haven't seen *anything* > that can be considered a true > innovation in Linux unless you consider being > DIFFICULT TO ADMIN an innovation.
Yeah, there's nothing like the ease of changing things like "HKEY_LOCAL_MACHINE....."
> Oh yes, eventually the open-source programmers > will steal the best ideas from > the other platforms, and after 2 years, get a > bug-free implementation, and
If Microsoft could get their bug cycle down to 2 years it'd have a positive effect on the entire planet. Funnily enough, Win98 is a bugfix pack for Win95, and it's *still* buggy as hell.
> claim it as their own (while denigrating MS and > Apple), but the fact remains, Metcalfe is > basically right.
> I mean, look at Apache, it's not even > multithreaded. Or GCC, it's based on
Apache != Linux, but I think you'll find the NT version of Apache working just fine as a multi-threaded server. There's also a current work-in-progress on dev.apache.org that has threads. Of course, it doesn't let people gain access to your box remotely, so you may consider it feature-poor.
> 10 year old ideas about CPUs and > global analysis on the C language. Or
Yet MS couldn't even come up with a C compiler for the Alpha, their compiler is Intel-centric and staying that way. DEC's compiler is of course better, but then again, it comes from.... oh darnit! There's that old Unix thing again.
> Emacs, an application that makes Office 2000 > look slim, but delivers 1/100th > of the functionality.
Nothing makes Office look slim. As for functionality, Emacs is an editor, not a word processor. I don't think you could buy a clue.
> The Linux community seems to be held together > mby a mutual hallucination > that you can hack together a clone of 30 year > old tech, call it innovative, and > have a bunch of teenagers worship a pile of shit > that is good for web servers, > but useless as a desktop.
Funnily enough, I've been using Linux as my primary desktop for about the last 7 years. Maybe you're just too stupid?
> I've been running Windows2000B3 on my PC for 3 > weeks now and it still hasn't crashed.
Wow! A whole 21 days! As a desktop with a single user! I'm impressed.
You know, I'll probably get moderated down for this, but my impression of you can be summed up in a single word.
In B&W up until I got the 5x7 I was shooting mostly Delta 100. I really love the actuance in PMK. The one time I tried TMX the negatives were too thin in PMK, but it was N+2. I've been meaning to try Microphen, esp. for Delta 3200, but PMK makes me happy. Now I'm shooting 5x7 it's either cutting down 8x10 or switching to Kodak as far as I can tell for T-grain films. I'm going through some HP5+ now, which stains *really* well in PMK and may make the issue moot if it comes out well. Not sure what I'll do in the long-run though.
I've read lots of good things about XTOL, and it's on my list of 'if I get sick of Pyro' things to try.
I've all but given up on 35mm. On the odd occasion that I need to shoot something I'm not willing to spend time moving in on I'll haul out the Nikon.
My snapshot camera these days is a 645, and I just got a Canham MQC 5x7. That's right, 35 *square inches* of film. Not to mention the fact tha the MQC is _engineered_.
If your film comes in a cassette you're probably playing with the wrong format;) Anyway, you gotta love a hobby where Schiempflug is a real word.
As for B&W and the comment about DIY, I do my B&W film developing in PMK - an awesome developer. Still using Ilford Multigrade for the paper, but I'm eyeing alternatives. I do my own E6 as well, and I've done a few Ilfochromes (used to be called Cibachrome).
If you're interesed in a good photo site, http://www.photo.net/photo rules. HP/UX backed by Oracle and AOLServer with TCL/TK.
> This is why I am not pleased that IBM has > noticed Linux. They talk the talk, but they > don't walk the walk! How much Open Source have > they contributed. Some public domain sample > code is all I've ever seen.
Like any other large company, IBM isn't a single entity. As far as this question goes though -
IBM Secure Mailer (AKA Postfix) Quite a bit of work on Apache A fair ammount of Java stuff I'm pretty sure the data visualization stuff advertised this week on/. was source code The PKIX reference implementation was IBM and Iris
I'm sure there's more, maybe you haven't been looking?
> Orange Book C2 ratings are explicitly defined as > being WITHOUT A NETWORK CONNECTION.
Which begs the question "Why didn't Microsoft go after a *Red Book* C2 evaluation?" That *does* include networking.
If you're touting yourself as a network OS, you should go after a network rating IMO. Of course, that means the TCB has to extend into the network layer - and C2 isn't exactly a high bar.
> Anything above B2 (including B2) does not allow > extra components added to the system > (i.e, 3rd-party software), because B2+ systems > have EVERYTHING hard-wired at the > BIOS (and I mean EVERYTHING).
This is patently untrue. You can read the specifications for B2 (red or orange book) and see that it is so. 3rd-party must simply be a part of the evaulation if it is to be a part of the Trusted Computing Base (TCB). For instance, TRW's DockmasterII code is under evaluation for B2 sitting on top of DG/UX's (also under evaluation) B2 version of Unix. B2 is indeed where things get serious, but it's all based on evaulation criteria in the orange or red books, and the TCB must enforce its rules, but the entire TCB of any modern computing system wouldn't fit into a BIOS.
The orange book is for stand-alone systems, the red book is for trusted networking implementations, and Data General is aiming for red book B2 - that means it's B2 with the network activated and even potentially sitting on the Internet.
For anyone interested in a Linux security project that aims to create a B-level security model, do a search for RSBAC, Mr. Ott has done a ton of good work in creating a real security model and protection mechanism for Linux at a level much higher than C2 which basicaly just means that you have to have an account to log in to the system.
investorrelation.sprintcom@mail.sprint.com
on
Rumours
·
· Score: 1
It would seem to me that as potential investors, the ideal thing to do would be to ask the investor relations department about Sprint's policy on the use of GNU, GPL and copylefted software. After all, we wouldn't want to invest in a company that didn't meet our investment objectives, now would we?
Slashdot Mirror
If you choose the "Select Individual Packages" option when installing you can pare it down pretty well. Especially if you don't let it install for failed dependancies.
I've only ever done a default install once, just to see what it was like. It didn't find enough room on the machine.
If you pick packages, you really need to have an idea of what's in them, for instance what -devel packages have which header files is a biggie. The help function to actually describe the package is significantly better than it used to be, but knowing what's in them is still a bit of trial and error.
After you're done, you can also look in the RPM database and deinstall anything extra that you don't like. I've never had difficulty fitting a reasonable X-less install on any old 486 machines I've found.
Paul
You're crying wolf. Linux and the associated softwre surrounding it are continually evolving. The kernel release schedule should make that patently obvious. RedHat is keeping as close to up as possible without losing some stability.
.98p11 SLS distribution, AIX for about the same ammount of time) I'm perfectly happy to upgrade my servers when they need it, no matter *which* OS or vendor I happen to use.
Yes, they've made mistakes in the past, so what?
You're free to create or use another distribution, fact is that RH's model is working for RedHat, why snivel about it?
Any company large enough to support AIX environments will have at least one person with enough of a clue to (a) wait for stable releases, and (b) FTP. RedHat's done a great deal over the last year to legitimize what we used to have to _sneak_ onto machines. The value in that alone is worth the rest of it to me in my daily job.
Having just gotten capital approval to replace some AIX boxen with both Linux and *BSD machines, and having played the patch/update game for quite a while in both environments (Linux since a
If you've paid for AIX and AIX support, you'd know there's no comparison in cost.
I actually tend to buy more CDs personally than we do at work because the release cycle isn't long enough to match my work maintenance cycle. But then I buy more FreeBSD CDs too. FTP installs work just fine from work too, multiple DS-3's make it satisfactory to do.
RedHat also makes updates available for the last couple of releases. Nobody's "forced" to upgrade for bugfixes if they're running a relatively new version-- even if they don't understand that they can apply the patches themselves and compile their own packages.
If you want to complain about people being uninformed, educate them, if you're just whining because RH is being successful, you're more than welcome to try to build your own brand.
RedHat isn't Linux. Linux isn't RedHat. Time you opened *your* eyes to that fact.
Paul
I take exception with the often parroted:
> Just look at OpenBSD. It's open source, yet it's > the most secure *nix OS on earth
While OpenBSD is well-done, and fairly secure, it's had exploits available in the past, and probably will have them in the future. I'd go as far as "Damn good security", but nowhere near "most secure." This isn't meant to malign the OBSD team, they've done a significant ammount of very good work.
Sun's Trusted Solaris and HP's Compartmented Mode Workstation (CMW) have a higher level of assurance. I'm not sure where Trusted Irix stands in the stack of things, but it's probably close to or on-par also.
I'd trust Data General's DG/UX- B2 feature set a *lot* more than I'd trust OpenBSD (indeed, NSA seems to be under the impression that it's good enough to house up to SECRET data and still be connected to the Internet- though it's FER seems overdue.)
Finally, I'd also probably put more stock in a properly administered version of Linux with RSBAC (http://www.compuniverse.com/rsbac/) than in OpenBSD because as good as dilligence and auditability are, role enforcement by a security process built into the kernel wins every time.
RSBAC (Ruleset Based Access Control) is an open source project that deserves scrutiny if you're interested in actual security. It's a work in progress that attempts to add B1-type features to the Linux kernel.
Paul
> What I'd really like to find is a way to run my > employer's software token in a *Nix environment > so that I don't have to use Windows to dial up
(a) ask the vendor about a Linux port - you never know what you might find out. My Security Dynamics salesdweeb just sent me a Linux executable that's either their client or server code - haven't had time to untar it and look at it yet. But asking for "Whatever you've got that runs on Linux" seems to have paid off with a few hundred K of gzipped tarball.
> to the office. Regrettably, they use a hardware > firewall device whose software clients only come
> in one flavor. Short of buying another computer
> and using my windows box for a proxy server, I
> have no way of connecting a more pleasant OS to
> the network where I work
Run NT or 98 in a VMWare window - if all you're using is a software challenge/response (Soft token) program, the VM shouldn't need to connect.
VMWare is $99 for home use, which is cheaper than a new machine. You can also run multiple VMs, so it's good for upgrade testing, testing across versions, etc.
If it has to connect, and you get to do PPP negotiation first, then just IP Masq the VM.
Paul
If it's running on something like DOCKMASTER II, it really doesn't need patching. DG/UX B2 is a network-level B2 (under eval, not done) that allows you to define untrusted paths coupled with the IP stack. So, even if you find an exploit, it's likely that the system is set up to not let you reach the level required to execute anything, and the httpd shouldn't be given things like access to a shell. Compartmented systems work fairly well for things like Web servers, nameservers and stuff like that. Data coming in from the Internet is always best set to the lowest level of trust. Web servers, etc. run at the path/user's MAC level/Capability set.
If you're remotely interested in the types of things you can do with OS-level security, do a search for RSBAC to see the types of princilples involved applied to a modern Linux kernel.
Paul
I saw something on the Park Police looking at or using cameras that mapped blood vessels in the face a while back. It's my guess that they're using something similar- that'd be much more accurate than a fingerprint scan and significantly harder to fake.
Paul
What about making ACs log in as AC and accept a session cookie to post? Valid expired AC session cookies could re-enable later AC access with a cookie exchange, you'd still have anonymous posting, but you'd also have a mechanism that would link multiple posts to a single AC in most instances. Anyone who was moderated down into slashdork status could be redirected to a "behave yourself" page if they still presented the same cookie. Multiple logins from the same IP as an AC within a timeframe could be slowed down like password timeouts to catch scripters - threshold settable per IP even.
> Comon, I dare you to name a few Linux-only > inventions of significant value.
The way Linux does syscalls, which was eventually added to Solaris due to the performance gains would be a good one. Running on systems from 8-bit to 64-bit processors seems to be a good thing that only NetBSD shares with any force. If it wasn't the first 64-bit clean OS, it was certainly one of them. Meanwhile, NT on my Alpha at work still runs in 32-bit mode. Fully-functional packet screening protection in a common kernel seems to have come out of the
Open Source community (though perhaps not Linux specificly.)
> Everything in the Linux kernel is a
> ripoff.
Bullshit. Most of the code was written from scratch. If you want to talk about processes and the kernel model, then nobody's made significant advances since kernel threading. For that matter, perhaps you can explain what benifits you expect to derrive from this "innovation" in NT? I see a bloated, semi-functional monolithic OS that's so slow that large parts of the user interface had to be moved into kernel space to make it usable on 300Mhz processors. Innovative that was! Maybe you like the innovation of a screen saver taking up 70% of a server's CPU?
> The point Metcalfe is trying to make is, other
> companies, like Be, NeXT, Taligent, Sun, Apple,
> even Microsoft, are atleast trying to do
> something different.
Most of whom are (a) failing, and (b) scrambling to continue the revenue generating mechanisms that keep them afloat. MS does a mediocre job in the desktop OS arena, where it really excells compared to everywhere else it's at. Apple does better, but doesn't have market share. Betamax syndrome.
> Linux is a very simplistic, naive, conservative > implementation of Unix. It's a
> monolithic kernel, it doesn't have the concept
> of a Yellow box like Mac OS
> X, it doesn't have asynchronous or streaming
> drivers like Win2k, it doesn't
> have a radical filesystem like Be, it doesn't
> have a great threading architecture
> like Solaris or Be, it's not distributed like
> Plan 9. It doesn't have formal
> verification, it doesn't have a virtual machine.
And yet every day it does its job just fine. Funnily enough Be doesn't seem to be moving into mission-critical positions. Funnily enough, Beowulf-class clusters seem to have rocketed Linux up the Top 500 supercomputer list without any significant OS-level distribution.
Try writing a driver for Linux then try to write one for NT. Even if you've never written a driver before, Linux will take a few days. Even if you've written drivers before, NT will take weeks.
> Basically, ZERO Innovation, written by a bunch
> of part-time hobbyists
> copying 30 year old designs with a few stolen
> ideas like sendfile()
Really interesting that if we took your belief (wrong though it is) that a bunch of part-time hobbyists seem to be challenging the halls of innovation and winning. Sun, IBM and SGI are all working on Linux support for their OS'. Microsoft is scrambling to try to protect and grow their server market while constantly losing ground to Linux. Good thing it's not a bunch of full time hobbyists if you hold MSFT stock, eh?
> Let's look at what commercial vendors have been > doing:
> o parallel pipeline window systems (years
> before X/Precision Insight)
Great market share there eh?
> o virtual machine / o-o architecture (Yellow
> Box, Java, NeXT, Taligent)
VMs were done well before Next and Taligent were ever companies. Perhaps you forget VM/CMS? As for O-O, it's slow and bloated. Smalltalk proved that case decades ago.
> o microkernel (NT, NeXT, Be)
NT is not a microkernel. The MkLinux port has a real microkernel. It's slow and inefficient compared to a monolithic kernel, hence the wishes of Mac-based Linux users to port the "normal" kernel.
> o fully threaded OS (Be)
pthreads work quite well for user-space programs, but kernel threads are happening.
> o component architecture (NT/COM)
COM sucks, HTH.
> o distributed components built in (DCOM/MTS)
DCOM sucks more, HTH.
> o directory service (Novell, NT)
Um, "So?"
> o plug and play architecture (Win95/98,and
> Win2k)
Linux has some plug-and-play features. NT doesn't. Generally it's not as robust or functional as MS would have wished for.
> o relational file system (Be)
Once again, "So?" Journaling file systems have a
great deal of value.
> o postscript based WYSIWYG display system
> (Display Postscript on the NeXT and Sun NeWS)
This is not an OS feature. Lyx works well as a WYSIWYG thing if you're into page layout though.
> The list just goes on. I haven't seen *anything* > that can be considered a true
> innovation in Linux unless you consider being
> DIFFICULT TO ADMIN an innovation.
Yeah, there's nothing like the ease of changing things like "HKEY_LOCAL_MACHINE....."
> Oh yes, eventually the open-source programmers
> will steal the best ideas from
> the other platforms, and after 2 years, get a
> bug-free implementation, and
If Microsoft could get their bug cycle down to 2 years it'd have a positive effect on the entire planet. Funnily enough, Win98 is a bugfix pack for Win95, and it's *still* buggy as hell.
> claim it as their own (while denigrating MS and > Apple), but the fact remains, Metcalfe is
> basically right.
> I mean, look at Apache, it's not even > multithreaded. Or GCC, it's based on
Apache != Linux, but I think you'll find the NT version of Apache working just fine as a multi-threaded server. There's also a current work-in-progress on dev.apache.org that has threads. Of course, it doesn't let people gain access to your box remotely, so you may consider it feature-poor.
> 10 year old ideas about CPUs and
> global analysis on the C language. Or
Yet MS couldn't even come up with a C compiler for the Alpha, their compiler is Intel-centric and staying that way. DEC's compiler is of course better, but then again, it comes from.... oh darnit! There's that old Unix thing again.
> Emacs, an application that makes Office 2000
> look slim, but delivers 1/100th
> of the functionality.
Nothing makes Office look slim. As for functionality, Emacs is an editor, not a word processor. I don't think you could buy a clue.
> The Linux community seems to be held together
> mby a mutual hallucination
> that you can hack together a clone of 30 year
> old tech, call it innovative, and
> have a bunch of teenagers worship a pile of shit > that is good for web servers,
> but useless as a desktop.
Funnily enough, I've been using Linux as my primary desktop for about the last 7 years. Maybe you're just too stupid?
> I've been running Windows2000B3 on my PC for 3
> weeks now and it still hasn't crashed.
Wow! A whole 21 days! As a desktop with a single user! I'm impressed.
You know, I'll probably get moderated down for this, but my impression of you can be summed up in a single word.
The word is:
Fuckwit.
In B&W up until I got the 5x7 I was shooting mostly Delta 100. I really love the actuance in PMK. The one time I tried TMX the negatives were too thin in PMK, but it was N+2. I've been meaning to try Microphen, esp. for Delta 3200, but PMK makes me happy. Now I'm shooting 5x7 it's either cutting down 8x10 or switching to Kodak as far as I can tell for T-grain films. I'm going through some HP5+ now, which stains *really* well in PMK and may make the issue moot if it comes out well. Not sure what I'll do in the long-run though.
I've read lots of good things about XTOL, and it's on my list of 'if I get sick of Pyro' things to try.
I use the liquid from the Formulary folks. I really love it especially with Delta 100.
>>When the ladies visit your house, a good book >>may set the mood.
> Here's a tip: Advanced Programming in the UNIX
> Environment doesn't seem to produce the
> desired results...
Perhaps you need to meet a better class of lady?
I've all but given up on 35mm. On the odd occasion that I need to shoot something I'm not willing to spend time moving in on I'll haul out the Nikon.
;) Anyway, you gotta love a hobby where Schiempflug is a real word.
My snapshot camera these days is a 645, and I just got a Canham MQC 5x7. That's right, 35 *square inches* of film. Not to mention the fact tha the MQC is _engineered_.
If your film comes in a cassette you're probably playing with the wrong format
As for B&W and the comment about DIY, I do my B&W film developing in PMK - an awesome developer. Still using Ilford Multigrade for the paper, but I'm eyeing alternatives. I do my own E6 as well, and I've done a few Ilfochromes (used to be called Cibachrome).
If you're interesed in a good photo site, http://www.photo.net/photo rules. HP/UX backed by Oracle and AOLServer with TCL/TK.
> This is why I am not pleased that IBM has
/. was source code
> noticed Linux. They talk the talk, but they
> don't walk the walk! How much Open Source have
> they contributed. Some public domain sample
> code is all I've ever seen.
Like any other large company, IBM isn't a single entity. As far as this question goes though -
IBM Secure Mailer (AKA Postfix)
Quite a bit of work on Apache
A fair ammount of Java stuff
I'm pretty sure the data visualization stuff advertised this week on
The PKIX reference implementation was IBM and Iris
I'm sure there's more, maybe you haven't been looking?
Some stuff is at http://www.alphaworks.ibm.com/
> Orange Book C2 ratings are explicitly defined as > being WITHOUT A NETWORK CONNECTION.
Which begs the question "Why didn't Microsoft go after a *Red Book* C2 evaluation?" That *does* include networking.
If you're touting yourself as a network OS, you should go after a network rating IMO. Of course, that means the TCB has to extend into the network layer - and C2 isn't exactly a high bar.
Look at:
http://www.compuniverse.com/rsbac/
Don't expect it to be tested though - that's expensive. It does raise the bar to B level though.
> Anything above B2 (including B2) does not allow > extra components added to the system
> (i.e, 3rd-party software), because B2+ systems
> have EVERYTHING hard-wired at the
> BIOS (and I mean EVERYTHING).
This is patently untrue. You can read the specifications for B2 (red or orange book) and see that it is so. 3rd-party must simply be a part of the evaulation if it is to be a part of the Trusted Computing Base (TCB). For instance, TRW's DockmasterII code is under evaluation for B2 sitting on top of DG/UX's (also under evaluation) B2 version of Unix. B2 is indeed where things get serious, but it's all based on evaulation criteria in the orange or red books, and the TCB must enforce its rules, but the entire TCB of any modern computing system wouldn't fit into a BIOS.
The orange book is for stand-alone systems, the red book is for trusted networking implementations, and Data General is aiming for red book B2 - that means it's B2 with the network activated and even potentially sitting on the Internet.
For anyone interested in a Linux security project that aims to create a B-level security model, do a search for RSBAC, Mr. Ott has done a ton of good work in creating a real security model and protection mechanism for Linux at a level much higher than C2 which basicaly just means that you have to have an account to log in to the system.
"MS DeWitt" Such a likely name for a lawyer...
Rat bastard!
;)
Any luck Intel will hit it with a backhoe
It would seem to me that as potential investors, the ideal thing to do would be to ask the investor relations department about Sprint's policy on the use of GNU, GPL and copylefted software. After all, we wouldn't want to invest in a company that didn't meet our investment objectives, now would we?
investorrelation.sprintcom@mail.sprint.com
Would seem to be one such place to query.