I suspect it's a little more complicated than that. Many of us have compiled our share of crypto source code, professionally and avocationally, can certainly disassemble things if we have to, but how many of us are mathematicians? Even if you can read the code, can you vet the math?
Much more importantly, how good is the key management? You can have the best algorithm in the world, but if you do sloppy key management, it's worse than useless, because you _think_ you're actually protected.
I've done security work, and I've watched people type PGP passphrases over unencrypted TELNET sessions. Bang. They're dead. Don't blame the crypto.
Until commercial key management options reach the levels of integrity that are necessary to avoid shooting ourselves in the foot, AND people learn how to use them properly, don't expect much.
Just wait. Even if the US lightens up on the algorithm export issues, watch how they handle advanced key management mechanisms. Should prove instructive.
See, I've been an OS/2 zealot for about the past 5 years, or whenever Warp came out. I loved it and used it as my home desktop of choice. I personally found the UI to be quite nice, and CONSISTENT. (Having been a Mac developer, that was important to me.)
I've basically switched to BeOS, with much soul-searching. Even with my paying IBM for quarterly updates, the support options were horrific, although I imagine the pricing is about the same as M$.
If OS/2 _really_ had Win32 interoperability, I might still be using it, but BeOS speed, flexibility and POSIX mostly-compliance are great things to me. It feels like the best of both worlds - If I want to port some UNIX hack, it's straightforward (EMX is fine, but only part of the picture.) If I want a nice UI, the BeOS UI rocks.
That having been said, my corporate web/mail/meeting maker server has been running OS/2 for 2 years without incident (security or otherwise) other than a power supply dying. I don't necessarily expect the BeOS to ever be that kind of platform.
I have to concur that the perception of OS/2 is exceedingly unfair. It's super-stable, and the virtualization is unparalleled, though I guess the VMware people are trying really hard to parallel it, as it were.
But..even getting past the fact that many or most/.ers seem to detest actually paying for an OS, IBM has done their best to make it hard to buy. Sure, ISV's are trying to take pieces of the new one and make a consumer accessible OS, but, if the vendor doesn't care about supporting me, I don't care about supporting it.
Don't you think there's an extreme danger of creeping featurism in your philosophy? I know it's heresy to say it, but having written software for a living several times in my life, sometimes you have to know when to STOP and "ship it."
I'm not suggesting that it's necessarily bad to allow groups of people to improve things, but I think that a truly inspired API and a group of clueful kernel developers can foster many or most of the "improvements" that you'd potentially see from an open source development effort.
I've used proprietary OS' in the past that Did The Right Thing so well, I haven't seen any successors really match them. Granted, most of them aren't in use now, but I'd be happy to just see us learn from the past. I have yet to see one OS that was a panacea for everything, and I don't think I ever will.
For example, if there were an OS that combined the job scheduling flexibility of TOPS-20, the self-diagnostic mechanisms of LISPM's, the security of Multics, and the UI of the Xerox Alto, you'd be ahead of most of what's out there.
Why don't you have such a thing? Because people don't learn from mistakes, and allow creeping featurism to distract them from their core goal.
On the not learning front, I count the Linux TCP/IP stack as a prime offender. It has come a heck of a long way, but slow start still doesn't seem to work right, and option handling is still coming along. Would it have been such a crime to learn from the BSD stack? I think not. The MS-Windows interface is another grand example - at least the Mac had human interface guidelines that were consistent.
On the creeping featurism front, I think that all the Unices are potential losers. Don't get me wrong, they're a damn sight better than most of what's out there, but does a _kernel_ REALLY need to be 2MB? GEOS managed in 120KB....and to think we give M$ a hard time for s/w bloat.
I think we sometimes miss the forest for the trees. A good proprietary OS is fine, if the API and the vendor support are decent. In much the same way I promised to pay for phone calls when I grew up, I'm willing to pay for an OS that is robust. That shouldn't threaten anyone who wants a free one.
Let's face it, there's more than a slight parallel to the Windoze/Mac dichotomy here.
Echostar apparently feels very strongly about maintaining control over enough aspects of the infrastructure to ensure a consistent, and hopefully satisfactory user experience. As a slightly later entrant, they had the ability to use a real standard (Is there such a thing?), whereas Hughes had to homebrew part of the protocol.
In any event, Echostar is more like Apple in this game, in that by providing both user hardware, programming services, and infrastructure, they can respond more quickly to their users, and probably make better margins as well.
I'm an E* customer, and I am generally pretty darn happy with the service. I have heard that some DTV/USSB users (probably not/. types) have had trouble with customer support becoming a ping-pong between the h/w vendor and the service vendors' support departments. I don't have that problem, since they're basically the same people for us DiSH heads.
That does not mean there aren't problems with support occasionally for E*, but between reading alt.dbs.echostar, the monthly customer chats, and their phone support, I've generally been able to fix what needed fixing, on the rare occasions something did.
I'm an old-guard DiSH network subscriber (I have a model 2000 in addition to my 5000 if that gives you any idea), and I love the service.
I tried to talk to them about getting access to one of the Broadlogic (spun off from Adaptec) ABA-1020 (IIRC) receiver boards, but I just got repeated requests to tell them my exact application. I'll probably run straight to Broadlogic and get a card for some other VSAT service, since Echostar won't let Broadlogic sell the cards directly, sigh.
If E* comes out with a server platform that I can use as a developer (was that the thing on the technical chat earlier this week? - I had to take a phone call.), I'll certainly consider using it.
At the risk of digressing, I for one, would cackle with glee. As a security person, I couldn't believe that MS wanted to spray native binaries at people through a web browser, and I still can't.
It's clear that some parties probably have an interest in AIM/ICQ being "standards" for chat mechanisms. I think the best thing anyone could do is write a Zephyr plug-in for Mozilla.
Slashdot Mirror
I suspect it's a little more complicated than that. Many of us have compiled our share of crypto source code, professionally and avocationally, can certainly disassemble things if we have to, but how many of us are mathematicians? Even if you can read the code, can you vet the math?
Much more importantly, how good is the key management? You can have the best algorithm in the world, but if you do sloppy key management, it's worse than useless, because you _think_ you're actually protected.
I've done security work, and I've watched people type PGP passphrases over unencrypted TELNET sessions. Bang. They're dead. Don't blame the crypto.
Until commercial key management options reach the levels of integrity that are necessary to avoid shooting ourselves in the foot, AND people learn how to use them properly, don't expect much.
Just wait. Even if the US lightens up on the algorithm export issues, watch how they handle advanced key management mechanisms. Should prove instructive.
Wow...Timur Tabi talking about BeOS.
/.ers seem to detest actually paying for an OS, IBM has done their best to make it hard to buy. Sure, ISV's are trying to take pieces of the new one and make a consumer accessible OS, but, if the vendor doesn't care about supporting me, I don't care about supporting it.
See, I've been an OS/2 zealot for about the past 5 years, or whenever Warp came out. I loved it and used it as my home desktop of choice. I personally found the UI to be quite nice, and CONSISTENT. (Having been a Mac developer, that was important to me.)
I've basically switched to BeOS, with much soul-searching. Even with my paying IBM for quarterly updates, the support options were horrific, although I imagine the pricing is about the same as M$.
If OS/2 _really_ had Win32 interoperability, I might still be using it, but BeOS speed, flexibility and POSIX mostly-compliance are great things to me. It feels like the best of both worlds - If I want to port some UNIX hack, it's straightforward (EMX is fine, but only part of the picture.) If I want a nice UI, the BeOS UI rocks.
That having been said, my corporate web/mail/meeting maker server has been running OS/2 for 2 years without incident (security or otherwise) other than a power supply dying. I don't necessarily expect the BeOS to ever be that kind of platform.
I have to concur that the perception of OS/2 is exceedingly unfair. It's super-stable, and the virtualization is unparalleled, though I guess the VMware people are trying really hard to parallel it, as it were.
But..even getting past the fact that many or most
Don't you think there's an extreme danger of creeping featurism in your philosophy? I know it's heresy to say it, but having written software for a living several times in my life, sometimes you have to know when to STOP and "ship it."
...and to think we give M$ a hard time for s/w bloat.
I'm not suggesting that it's necessarily bad to allow groups of people to improve things, but I think that a truly inspired API and a group of clueful kernel developers can foster many or most of the "improvements" that you'd potentially see from an open source development effort.
I've used proprietary OS' in the past that Did The Right Thing so well, I haven't seen any successors really match them. Granted, most of them aren't in use now, but I'd be happy to just see us learn from the past. I have yet to see one OS that was a panacea for everything, and I don't think I ever will.
For example, if there were an OS that combined the job scheduling flexibility of TOPS-20, the self-diagnostic mechanisms of LISPM's, the security of Multics, and the UI of the Xerox Alto, you'd be ahead of most of what's out there.
Why don't you have such a thing? Because people don't learn from mistakes, and allow creeping featurism to distract them from their core goal.
On the not learning front, I count the Linux TCP/IP stack as a prime offender. It has come a heck of a long way, but slow start still doesn't seem to work right, and option handling is still coming along. Would it have been such a crime to learn from the BSD stack? I think not. The MS-Windows interface is another grand example - at least the Mac had human interface guidelines that were consistent.
On the creeping featurism front, I think that all the Unices are potential losers. Don't get me wrong, they're a damn sight better than most of what's out there, but does a _kernel_ REALLY need to be 2MB? GEOS managed in 120KB.
I think we sometimes miss the forest for the trees. A good proprietary OS is fine, if the API and the vendor support are decent. In much the same way I promised to pay for phone calls when I grew up, I'm willing to pay for an OS that is robust. That shouldn't threaten anyone who wants a free one.
P.S.: This was written on a BeOS/Intel machine.
Let's face it, there's more than a slight parallel to the Windoze/Mac dichotomy here.
/. types) have had trouble with customer support becoming a ping-pong between the h/w vendor and the service vendors' support departments. I don't have that problem, since they're basically the same people for us DiSH heads.
Echostar apparently feels very strongly about maintaining control over enough aspects of the infrastructure to ensure a consistent, and hopefully satisfactory user experience. As a slightly later entrant, they had the ability to use a real standard (Is there such a thing?), whereas Hughes had to homebrew part of the protocol.
In any event, Echostar is more like Apple in this game, in that by providing both user hardware, programming services, and infrastructure, they can respond more quickly to their users, and probably make better margins as well.
I'm an E* customer, and I am generally pretty darn happy with the service. I have heard that some DTV/USSB users (probably not
That does not mean there aren't problems with support occasionally for E*, but between reading alt.dbs.echostar, the monthly customer chats, and their phone support, I've generally been able to fix what needed fixing, on the rare occasions something did.
I'm an old-guard DiSH network subscriber (I have a model 2000 in addition to my 5000 if that gives you any idea), and I love the service.
I tried to talk to them about getting access to one of the Broadlogic (spun off from Adaptec) ABA-1020 (IIRC) receiver boards, but I just got repeated requests to tell them my exact application. I'll probably run straight to Broadlogic and get a card for some other VSAT service, since Echostar won't let Broadlogic sell the cards directly, sigh.
If E* comes out with a server platform that I can use as a developer (was that the thing on the technical chat earlier this week? - I had to take a phone call.), I'll certainly consider using it.
You forgot one of the oldest and most time-honored:
"One who makes furniture with an axe."
At the risk of digressing, I for one, would cackle with glee. As a security person, I couldn't believe that MS wanted to spray native binaries at people through a web browser, and I still can't.
It's clear that some parties probably have an interest in AIM/ICQ being "standards" for chat mechanisms. I think the best thing anyone could do is write a Zephyr plug-in for Mozilla.