IANAGCCD (I am NOT a GCC developer;-))) but AFAIK we should all thanks to RedHat for putting lot of "people-time" (read - money) into GCC and making constant improvements of it!
Speaking as somebody from post-communist block, the problem with socialism in purest form (communism) is that it looks very nice on paper, but in practise there these main problems: - "distributing money" through taxes (in) and social and other support (out) is the very most un-effective way (LOT of money is "lost" and whole process tends to be very very expensive (lot of buerocracy)) - (some) people tends to be greedy (they try to avoid to share) etc. etc.
But this is all about politics - I think that applying "real world" scheme to the SW (and IP) is just plainly wrong - that is totaly different "ecosystem".
It's all question about price... If you can get A4-A3 size for 10$, then it has potential. If it's 100$, then it'll not be wide-spread and if 1000$, then it'll never go to production. Of course I'm NOT talking about price for some samples and I'm talking just about price of the "paper" + some controller, not about storage, CPU etc. etc.
I just don't see one, the most fundamental, question in all interview. I don't worry about climber construction or powering them (it's after all "just" engineering - even if powering means, that you'll put very very small reactor on the climber and restrict it going only from 1000 Km and higher and for 0-1000 you'll use chemical rockets) - BUT (!) AFAIK the material is problem! I've read somewhere, that the strongest nanotube ever produced is still only 50% of necessary strength - and THAT'S a LONG way to go! (you can't use just 100% necessary strength - you need more for safety - something like 130-150%!)
AFAIK it was not just desert but there were lot of obstacles on the road. And remember that the main mid-term target of this competition is to create the autonomous supply truck for army - and there you just don't care about the speed (the urgent cargo can be sent always by heli or drop-down).
Now only if Windows can correctly boot on completely different box... Author probably never tried to take his Windows XP disk and boot in different box with different mainboard, video and network card...
I was on some conference (Samba eXPerience) and there was talk about AD Kerberos modification - I just remember one thing ("executive summary") - Microsoft's modifications to Kerberos are very very usefull and they are done in quite "clean" and "kerberos"-way and are necessary for using Kerberos in large scale networks (you could use Kerberos without them but you'll lack some functionality). Please no flame - of course that MS fucked up the way of implementing this (they should made the changes more open way and discuss it more with Kerberos community).
Yeah, you're right, but it tells something about "code culture" when they don't do input validation even in such a critical (from bussiness/money point of view) module.
That the fingerprint recognition is quite simple algorithm (compared t "generall computer image understanding" which is very very complex where you've to deal with colors, perspective, "intuition" etc.)
Yeah - that's why I'm saying that there is nothing on the horizont. If something should become more widely usable in 2 years, we'd be seeing some beta realease, flame wars on/. why this is better/worse then PAM, people pushing this into Fedora Core 5 etc. right NOW.
Of course that if you write something like that you'll have some library with your "kind-of-API" (more or less public and stable). I just wanted to say, that there is no need to write something that will replace PAM just to get biometric API - and I don't think that IBM has done it.
Less lock in, since when the next generation of PAM killer comes along, the switch will be much easier.
That's stupid. There is nothing like "PAM killer" on the horizont in next 1-2 years! And there is no need for it - AFAIK PAM architecture is very clever and there are none "system design limitations" (but I'm NOT PAM expert - if I'm wrong, please correct me!)
Better portability to systems that don't use PAM. QNx, ReactOS, Windows, MacOS the world is a big place...
AFAIK MacOS is using PAM (or not?). And writing new API means that you've to transfer (and integrate it into existing) Windows/QNX... OS. The effort is much bigger then having "proprietary" library and just port it to Windows native login API/Linux PAM/...
More uses for the software. Maybe you can use this fingerprinter together with a Firefox plugin to slightly increse the security of your bank transactions?
WRONG! Just make FireFox PAM plugin and voila - you can use your "PIN pad" (if it has PAM plugin), fingerprint/face/voice/DNA/... recognition (just by having PAM plugin for this) out of box!
AFAIK not - fingerprint is just "convert black&white image to curves, find markers (like end of "line", join of 2 lines etc.) and save relative position of these markers. In fact fingerprint "image" is usually a few 10s of bytes!
PAM is really great thing - you can even have "plaintext" passwords in *SQL database or whatever - so there is no need to change hash or anything. IIRC I've seen some biometric Linux solutions (using PAM) on some CeBIT show...
Slashdot Mirror
I already know 1st step "HowTo run Linux on your XBox 360" - it's:
;-)
1, Attach your XBox on a string so it can be c00l enough
IANAGCCD (I am NOT a GCC developer ;-))) but AFAIK we should all thanks to RedHat for putting lot of "people-time" (read - money) into GCC and making constant improvements of it!
Speaking as somebody from post-communist block, the problem with socialism in purest form (communism) is that it looks very nice on paper, but in practise there these main problems:
- "distributing money" through taxes (in) and social and other support (out) is the very most un-effective way (LOT of money is "lost" and whole process tends to be very very expensive (lot of buerocracy))
- (some) people tends to be greedy (they try to avoid to share)
etc. etc.
But this is all about politics - I think that applying "real world" scheme to the SW (and IP) is just plainly wrong - that is totaly different "ecosystem".
What's so difficult to write "yum install application_name"? (and yes - there is GUI also - "yumex").
It's all question about price... If you can get A4-A3 size for 10$, then it has potential. If it's 100$, then it'll not be wide-spread and if 1000$, then it'll never go to production. Of course I'm NOT talking about price for some samples and I'm talking just about price of the "paper" + some controller, not about storage, CPU etc. etc.
I just don't see one, the most fundamental, question in all interview. I don't worry about climber construction or powering them (it's after all "just" engineering - even if powering means, that you'll put very very small reactor on the climber and restrict it going only from 1000 Km and higher and for 0-1000 you'll use chemical rockets) - BUT (!) AFAIK the material is problem! I've read somewhere, that the strongest nanotube ever produced is still only 50% of necessary strength - and THAT'S a LONG way to go! (you can't use just 100% necessary strength - you need more for safety - something like 130-150%!)
Have you seen Mac OS X last years? Yeah, it's UNIX inside ;-)
1, Why? Why store binary data in registry or INI files? And you can - just do UUENCODE / Base64 code and you can store it there
2, Why? Who says that? Why can't 1 INI be used with 100s of apps? They just need to know, where the INI is stored (just like knowing registry key)
3, You say that in UNIX is not multiuser? And UNIX is using INI-like-files for something like 30 years and it just works...
AFAIK it was not just desert but there were lot of obstacles on the road. And remember that the main mid-term target of this competition is to create the autonomous supply truck for army - and there you just don't care about the speed (the urgent cargo can be sent always by heli or drop-down).
Now only if Windows can correctly boot on completely different box... Author probably never tried to take his Windows XP disk and boot in different box with different mainboard, video and network card...
I was on some conference (Samba eXPerience) and there was talk about AD Kerberos modification - I just remember one thing ("executive summary") - Microsoft's modifications to Kerberos are very very usefull and they are done in quite "clean" and "kerberos"-way and are necessary for using Kerberos in large scale networks (you could use Kerberos without them but you'll lack some functionality). Please no flame - of course that MS fucked up the way of implementing this (they should made the changes more open way and discuss it more with Kerberos community).
Next programming model is right here
Yeah, you're right, but it tells something about "code culture" when they don't do input validation even in such a critical (from bussiness/money point of view) module.
Where can I pick up sarcasm? Somebody is givin' it for free? .-)
So that's what Microsoft did with the 5.8 B$ research budget - looking for new Windows release name :-)
That the fingerprint recognition is quite simple algorithm (compared t "generall computer image understanding" which is very very complex where you've to deal with colors, perspective, "intuition" etc.)
Yeah - that's why I'm saying that there is nothing on the horizont. If something should become more widely usable in 2 years, we'd be seeing some beta realease, flame wars on /. why this is better/worse then PAM, people pushing this into Fedora Core 5 etc. right NOW.
Of course that if you write something like that you'll have some library with your "kind-of-API" (more or less public and stable). I just wanted to say, that there is no need to write something that will replace PAM just to get biometric API - and I don't think that IBM has done it.
Less lock in, since when the next generation of PAM killer comes along, the switch will be much easier.
That's stupid. There is nothing like "PAM killer" on the horizont in next 1-2 years! And there is no need for it - AFAIK PAM architecture is very clever and there are none "system design limitations" (but I'm NOT PAM expert - if I'm wrong, please correct me!)
Better portability to systems that don't use PAM. QNx, ReactOS, Windows, MacOS the world is a big place...
AFAIK MacOS is using PAM (or not?). And writing new API means that you've to transfer (and integrate it into existing) Windows/QNX... OS. The effort is much bigger then having "proprietary" library and just port it to Windows native login API/Linux PAM/...
More uses for the software. Maybe you can use this fingerprinter together with a Firefox plugin to slightly increse the security of your bank transactions?
WRONG! Just make FireFox PAM plugin and voila - you can use your "PIN pad" (if it has PAM plugin), fingerprint/face/voice/DNA/... recognition (just by having PAM plugin for this) out of box!
AFAIK not - fingerprint is just "convert black&white image to curves, find markers (like end of "line", join of 2 lines etc.) and save relative position of these markers. In fact fingerprint "image" is usually a few 10s of bytes!
Of course having plaintext passwords is braindead stupid. I've said it to show, that PAM doesn' neccessary imply passowrd hashing.
PAM is really great thing - you can even have "plaintext" passwords in *SQL database or whatever - so there is no need to change hash or anything. IIRC I've seen some biometric Linux solutions (using PAM) on some CeBIT show...
What'd you think about airlines, where you have to rent a parachute/safe-boat as extra? ;-)
Yeah - but step 3. == step 4. ;-)
1, Sell OS which enables viruses to spread very easily
2, Create AV SW
3, Profit on selling AV SW AND new OS updates! Muhahaha...
Jeez, we're screwed...