> "Opps, shit did I link with something that requires I release the source".
that can't happen. all that can happen is "sorry, you can't use this unless your code is under a compatible license". and if that is the case usualy you say you are sorry, drop using that library/whatever and you are fine. worst case: you are forced to take your software off the market till the situation is fixed (wouldn't affect your windows or mac os X version unless you use the same library there too), and in theory someone could sue you for compensation of damages. But all the cases I followed that never happened.
don't let the fud get to you. some people claim if you combine A with B then A is suddenly under the same license as B. that is not true. but: you are not allowed to combine A and B unless their licenses are compatible. and if you make a mistake usually it is good enought to stop combining A and B or find some other compromise.
without investing too much time you can figure out which combinations are ok. don't worry too much, take come basic care and you will be fine.
I have servers running xen fine. but the situation with xen bothers me: documentation is bad, support on mailing list nearly dead, and xen always uses some old outdated linux kernel as base, so I never know if recent security updates made it into the xen kernel. So I wonder: is any of the alternatives ready to replace xen servers? kvm or lguest? I'm running linux below linux without hardware paravirtualisation support. or what about virtualbox - would be a much better vmware alternative.
on the other side xen absolutely rocks when it comes to integration. I don't need to think about the xen server in domU on my laptop - I stop, start and reboot my laptop as I see and the xen domain is always fine with it. not sure if the alternatives have reached that level yet.
2-3 weeks of development where changes go in, 9-10 weeks of testing and bug fixing. that is stable. you can't get a.0 release more stable than it is after 9-10 weeks of testing. if a 3 month cycle for new releases is not good enough, you have these options:
* use 2.6.16.* it has been maintained for about 18 months and still is
* use the kernel of your linux distribution. they typically maintain it for about 1-3 years, longer for the enterprise edition.
also remember that you don't need to switch to the latest version, no one is forcing you. a.0 release is only as good as the developer with their resources for testing can get it. the status is called "banana green" and it now need time with real users to mature.
as for the features you mentioned, they are additions, not replacement. having two implementations of something next to each other behind the same generic interface allows early testing and smooth migration. it doesn't harm the existing implementation or makes it less stable. simply not turn them on and you won't run into problems.
if you run a anti spam filter, it is your job to make sure your data is accurate. but if you think your users would pressure some admin so they get back to you, that is keeping mails hostage and not an acceptable practice.
if you do that, it is not part of the solution, it is part of the problem.
browser plugin for personality managemant please
on
The Case for OpenID
·
· Score: 1
well, can this help me to create a number of fake users (e.g. for all those stupid "please register" web sites), and help me to manage who site gets which personality. I would really prefer if I could thus decrease the number of sites that know me, and instead use throw away identities for "free downloads" and stuff like that.
It never worked as well as CISC, even though lots and lots of people tried to get it to work as well. So currently: CISC won. Until the next guy comes aroung with an idea for a much better RISC processor (and maybe fails like all of them before).
aladdin is well known for smart cards and usb crypto dongles with smart cards inside for a long while. so it won't be that easy. but as usual if you can find the "if (dongle == OK)" command in the code, then you can create a nocd version - not much new here.
tcpa has more interesting security properties like remote attestation. combining dongle and cd into one device - I fail to see how that is interesting.
computer technicans can make a bg impact by testing if some hardware works. and with that I mean: works for simple, unexperienced first time linux users. get those latest ubuntu/suse/fedora core previews, and give it a in detailed testing. you know how to use dmidecode to see what should work, you know the main board manual and all the whistles and bells that should be available (e.g. sensors, special keys on notebook keyboards, etc.). do a real test and give feedback to the distributions detailed enough, so they can get all of that working out of the box without any changes necessary..
linux doesn't fail on the first impression, and that is the status for a few years now. but once people use linux for a while, they sometimes stop liking it if some small problem pops up and is not easily solved. like notebooks not properly hibernating, serial ports not working, or some other issue.
as computer technican no one expects you to write that new kernel driver for some hardware, or financial software people might want for linux. but you can help where your expertise is and where help is needed: giving hardware a 100% check with even the smallest feature, and note down how to get them working (or that they don't work - even this is very valuable, so people don't waste hours to try). and as linux enthusiast you can not only test the released version of your preferred distribution, but also some pre-release. the final release always gets tons of users finding all kinds of problems, but only testers of pre-releases can find problems and report them with fixes in time to improve the quality of the final release.
doing an interview is about asking questions and letting the person you interview talk. I get annoyed a lot by stuff like this where the interviewer comments all the time or talks about his own agenda rather than giving the spotlight to the person interviewed.
drivers and kernel core are combined, so they can be managed as one. look at external/unmerged drivers: everytime the kernel core is improved they need to be updated and usualy lack behind. while in kernel drivers are updated at the same time so the kernel as combination of core plus driver is in harmony.
one alternative would be a stable driver api. that meant no improvements to drivers at all, which is not a good idea. or have several versions of such an api in parallel. look at windows and other operating systems - it is a maintenance nightmare to maintain several API in parallel. such an approach leads to much more code and much more bugs in the real world.
linux is a first class operating system, no alternative has so much hardware support, both cpu plattforms and drivers. it only works because all of it is combined in one source tree, and can be developed and improved in parallel.
and for runtime - simply use a modular kernel, it will load only the drivers you need. having extra drivers on hard disk is not a big problem (unless you care for those 80MB of modules, but then you are in an embedded world where you have much tighter requirements anyway).
wait actually 2.7 was 2.6.16-rc* and 2.8 was 2.6.16 and 2.6.17-rc* was 2.9 so 2.6.17 is 2.10.
if you do not like stable and well tested released every 3 months with lots of improvements, then you can simply use your distributions kernel, they maintain one kernel for years.
dsl (2mbit): 20euro. flatrate included - not traffic charges. the telephone line is also already included, no extra charges. if you want to phone - these days you would use voice over ip. free account already included, but with 1ct/min costs for POTS connections (or 10 euro per month for flatrate). dsl modem / router / voip gateway also included for free. tv? why not use a digital tv receiver? no extra charges (unless you want pay tv subscriptions, but most people are fine without).
looks like usa did fall quite behind? on the other side who are we to say that? germany isn't that good either, take a look at japan, south korea, sweden, finland, denmark,...
also available on windows?
on
Vim 7 Released
·
· Score: 1
who is planning that? I would like to get specs etc. as fast as possible or be part of the discussion on how the card will be. Countries like Estonia and Belgium did choose OpenSC open source software as basis for using their cards, and that would be a good choice for Australia, too. (Of course I'm biased being on of the OpenSC developers:)
Every three years I look at all the latest shiny hardware and write down the nicest machine I can think off, totaly bleeding edge, highest capacity and everything, but still something I think that might work. Then I put that paper aside.
6 months later those components are widely available at a reasonable price, drivers are stable and I can afford it, so I buy a new machine.
Having a new additional machine allows me to slowly move over to it, then take the three old one once I no longer need it and reinstall it as server, so the 6 year old can slowly retire. Works fine for me, not much hardware trouble, and still reasonable costs.
but that means usually a difficult choice: often there is an open source driver, but it is not fast. On the other hand the super fast optimized proprietory driver is a maintenance nightmare: you need to update it each time you update the kernel or Xorg and keep all those parts in sync.
Closed source binary drivers do not need to go away. But currently they do everything in the closed binary only part. Please change that to only do those parts in binary only that are real important (like all those secret tricks to get fast. And then open the rest - all that boring code that switches graphic modes and checks what monitor is attached and turns of the backlight if I close my laptop. There is no reason to keep those simple parts secret, and opening them would reduce the driver to a much smaller size, most likely move the proprietory parts into userspace / libGL and Xorg.
That would be a good step, make maintenance of kernel and Xorg and friends easier, and still allow NVidia and ATI to keep most of there secrets.
Come on ATI and NVidia, you can do it! Divide your graphic drivers into an kernel part and a user space part and open source the kernel part. There are no big trade secrets in switich a graphic card from mode A to mode B or finding out how fast it is clocked, what kind of monitor is attached and all that.
If you absolutely want to keep a proprietory binary only core under a closed licensed, that is fine. But please limit it to the real important stuff, like those mega tuned GL libraries and 3d accelleration, and make the basic parts of your driver open source. That way kernel and big parts of X11 and libs could be open source, and you could have additional, optional parts for X11 for better accelleration and stuff like that.
Well, ATI is already a good step into that direction, the radeon driver works pretty well and is only half as fast as the fglrx driver when it comes to video play back. Now please look at your code, merge all the non important parts with radeon, and reduce the fglrx driver to some user space binary blob. The kernel parts can't be that important, if radeon driver gets nearly everything right and fast.
Re:Webmail is a technological step backwards
on
Gmail vs Pine
·
· Score: 1
nah. think about email encryption/decryption, that is far too heavy on the cpu, your 386SX won't do it nicely. so lets move the encryption/decryption to a central server.
the crimimal part is not that bad - most prosecutors will simply drop small cases as not important, so there is little change. and doing private copies for yourself is techincaly not legall, but practical not enforceable.
the bad part is this: under the new law internet providers will have to answer to the media industry. so far we had a real data privacy act, i.e. the policy / prosecutors had to ask the internet providers for data, and then the media industry could have a look (and use it to start a civil law case). too bad the justice department gave in and wants to save the prosecutors some work. as side effect there is no way to stop the media industry from getting that data, there is no control, no one to stop them from overusing that power etc. lets hope the supreme court will sack that part of the law as conflicting with the german bill of rights (grundgesetz).
note: so far there is only a consensus within the governemnt and the lobby groups. german bundestag does not have much power - most politicians simply do what their party tells them to do, as they have next to no power without the party - but there is still hope they resist and successfully demand changes before passing that proposed new law.
did you notice how the first half talks about movies and the fear of piracy, but the second half only mentions the music industry not making as much money as they used to?
well, the german movie industry has their own association with a web site at http://www.bvv-medien.de/, and despide a very, very aggressive anti-consumer anti piracy campain, they still more than doubled their revenue in the last five years: 860 mio euro in 1999 vs. 1747 mio in 2004.
I guess noone of the german movie industry will read this, but: if I'm in cinema and about some movie, and you want to show some ad to me, it should start with the word "Danke" (thanks). After all I already paid for the movie ticket. Instead they show some anti piracy ad with people send into jail and about to be raped or similar stuff.
so far the ministry of justice has a proposal for a changed law. that doesn't mean the bundestag will accept that proposal without changes. at least joerg tauss of spd said there will be further long discussions, and no simply voting on it. so lets hope for the best.
nor is Feburary 30th. But german banks charge interest for the 30th of February. all the time. legal. even approved by the highest court. (something about exponential mathematics being too hard for the average citizen, so calculating with 360 days a year and linear interest is better).
You need to be more flexible if you want to target customers like german banks as well.
I tried to - and these bugs annoy me:
* zip files: the build in stuff does not work, double clicking on a zip file creates a new window of
a process that hangs. you need to kill it using task manager. 7zip&co to the rescue.
* "security" implemented badly: everytime anything gets executed you get a prior popup where you need
to allow that. for example when I want to run the control panel, I get that popup first. or clicking
on anythign within the control panel. even worse are the details: they tell me that a dll is about
to be run as program. which dll? which application triggered that? of course that information is
missing. and of course, vista does not remember my choice. next time I run the same application, I
get the same popup. stupid.
* slow as hell, sometimes. also sometimes astonishing fast. a very different mix, it takes time to get
used to it. he slow parts of course are annoying, and the faster parts don't cancel out that effect.
* looks like mac os X. sure, some people like it, I prefer kde. and all the parts on mac os X look and
feel that I hate are now also implemented in vista. great.
* auto-run still active: autorun for cds was a bad idea. the sony/drm/rootkit story should have told them
at last. but they didn't learn it enough, so now you get a popup whether or not to run some autorun
(or some other software). kde has the same popups, they are a bad idea like almost every popup.
animeate some icon if you want or something, but don't move some software in the front.
* graphics don't work well - yet. the nvidia driver shipped with it doesn't work well, the result has
strange graphic effects that noone wanted (I tried only WoW, but I guess other apps will have them too).
nvidia has a driver for vista 32bit! but it doesn't install - finds no supported hardware. crap.
* sound problems? the driver works fine, the test tool works fine, but applications like WoW stay silent.
maybe the vista audio api broke and is no longer compatible?
* doesn't shut down. the start menu has a button that looks like it will shut down and turn of the
machine, and I was impressed how fast it is. but it does not do that, vista only suspends to disk.
* and of course for everyone (including me) to judge a product by a beta 2 version. oops, it is not beta2,
it is only community technlogy preview with buils 5270 - guess microsofts own QA department knows well
that while some parts have already the "beta 2" label, it could be called at best "alpha 2" if you take
the (missing) usability into account.
Slashdot Mirror
> "Opps, shit did I link with something that requires I release the source".
that can't happen. all that can happen is "sorry, you can't use this unless your code is under a compatible license". and if that is the case usualy
you say you are sorry, drop using that library/whatever and you are fine. worst case: you are forced to take your software off the market till the situation
is fixed (wouldn't affect your windows or mac os X version unless you use the same library there too), and in theory someone could sue you for compensation
of damages. But all the cases I followed that never happened.
don't let the fud get to you. some people claim if you combine A with B then A is suddenly under the same license as B. that is not true. but: you are not
allowed to combine A and B unless their licenses are compatible. and if you make a mistake usually it is good enought to stop combining A and B or find
some other compromise.
without investing too much time you can figure out which combinations are ok. don't worry too much, take come basic care and you will be fine.
I have servers running xen fine. but the situation with xen bothers me: documentation is bad, support on mailing list nearly dead, and xen always uses some old outdated linux kernel as base, so I never know if recent security updates made it into the xen kernel. So I wonder: is any of the alternatives ready to replace xen servers? kvm or lguest?
I'm running linux below linux without hardware paravirtualisation support. or what about virtualbox - would be a much better vmware alternative.
on the other side xen absolutely rocks when it comes to integration. I don't need to think about the xen server in domU on my laptop - I stop, start and reboot my laptop as I see and the xen domain is always fine with it. not sure if the alternatives have reached that level yet.
2-3 weeks of development where changes go in, 9-10 weeks of testing and bug fixing. .0 release more stable than it is after 9-10 weeks of testing.
.0 release is only as good as the developer with their resources for testing can get it.
that is stable. you can't get a
if a 3 month cycle for new releases is not good enough, you have these options:
* use 2.6.16.* it has been maintained for about 18 months and still is
* use the kernel of your linux distribution. they typically maintain it for about 1-3 years, longer for the enterprise edition.
also remember that you don't need to switch to the latest version, no one is forcing you.
a
the status is called "banana green" and it now need time with real users to mature.
as for the features you mentioned, they are additions, not replacement. having two implementations of something next to each other
behind the same generic interface allows early testing and smooth migration. it doesn't harm the existing implementation or makes
it less stable. simply not turn them on and you won't run into problems.
if you run a anti spam filter, it is your job to make sure your data is accurate.
but if you think your users would pressure some admin so they get back to you,
that is keeping mails hostage and not an acceptable practice.
if you do that, it is not part of the solution, it is part of the problem.
well, can this help me to create a number of fake users (e.g. for all those stupid "please register" web sites), and help me to manage
who site gets which personality. I would really prefer if I could thus decrease the number of sites that know me, and instead use throw
away identities for "free downloads" and stuff like that.
It never worked as well as CISC, even though lots and lots of people
tried to get it to work as well. So currently: CISC won. Until the
next guy comes aroung with an idea for a much better RISC processor
(and maybe fails like all of them before).
aladdin is well known for smart cards and usb crypto dongles with smart cards inside for a long while.
so it won't be that easy. but as usual if you can find the "if (dongle == OK)" command in the code,
then you can create a nocd version - not much new here.
tcpa has more interesting security properties like remote attestation.
combining dongle and cd into one device - I fail to see how that is interesting.
computer technicans can make a bg impact by testing if some hardware works. and with that I mean: works for simple, unexperienced first time linux users.
get those latest ubuntu/suse/fedora core previews, and give it a in detailed testing. you know how to use dmidecode to see what should work, you know the
main board manual and all the whistles and bells that should be available (e.g. sensors, special keys on notebook keyboards, etc.). do a real test and give
feedback to the distributions detailed enough, so they can get all of that working out of the box without any changes necessary..
linux doesn't fail on the first impression, and that is the status for a few years now. but once people use linux for a while, they sometimes stop liking
it if some small problem pops up and is not easily solved. like notebooks not properly hibernating, serial ports not working, or some other issue.
as computer technican no one expects you to write that new kernel driver for some hardware, or financial software people might want for linux. but you can
help where your expertise is and where help is needed: giving hardware a 100% check with even the smallest feature, and note down how to get them working
(or that they don't work - even this is very valuable, so people don't waste hours to try). and as linux enthusiast you can not only test the released
version of your preferred distribution, but also some pre-release. the final release always gets tons of users finding all kinds of problems, but only testers
of pre-releases can find problems and report them with fixes in time to improve the quality of the final release.
alsa is much better, no need to use the old open sound system. or as api even suggest some higher level api used by both gnome and kde.
doing an interview is about asking questions and letting the person you interview talk.
I get annoyed a lot by stuff like this where the interviewer comments all the time or
talks about his own agenda rather than giving the spotlight to the person interviewed.
drivers and kernel core are combined, so they can be managed as one.
look at external/unmerged drivers: everytime the kernel core is improved they
need to be updated and usualy lack behind. while in kernel drivers are updated
at the same time so the kernel as combination of core plus driver is in harmony.
one alternative would be a stable driver api. that meant no improvements to drivers
at all, which is not a good idea. or have several versions of such an api in parallel.
look at windows and other operating systems - it is a maintenance nightmare to maintain
several API in parallel. such an approach leads to much more code and much more bugs
in the real world.
linux is a first class operating system, no alternative has so much hardware support,
both cpu plattforms and drivers. it only works because all of it is combined in one
source tree, and can be developed and improved in parallel.
and for runtime - simply use a modular kernel, it will load only the drivers you need.
having extra drivers on hard disk is not a big problem (unless you care for those 80MB
of modules, but then you are in an embedded world where you have much tighter requirements
anyway).
2.7 was 2.6.17-rc*, so this is 2.8.
wait actually 2.7 was 2.6.16-rc* and 2.8 was 2.6.16 and 2.6.17-rc* was 2.9 so 2.6.17 is 2.10.
if you do not like stable and well tested released every 3 months with lots of improvements,
then you can simply use your distributions kernel, they maintain one kernel for years.
dsl (2mbit): 20euro. flatrate included - not traffic charges. the telephone line is also already included, no extra charges.
...
if you want to phone - these days you would use voice over ip. free account already included, but with 1ct/min costs for POTS
connections (or 10 euro per month for flatrate). dsl modem / router / voip gateway also included for free. tv? why not use a
digital tv receiver? no extra charges (unless you want pay tv subscriptions, but most people are fine without).
looks like usa did fall quite behind? on the other side who are we to say that? germany isn't that good either, take a look
at japan, south korea, sweden, finland, denmark,
is microsoft still shipping edlin?
who is planning that? I would like to get specs etc. as fast as possible :)
or be part of the discussion on how the card will be. Countries like
Estonia and Belgium did choose OpenSC open source software as basis
for using their cards, and that would be a good choice for Australia, too.
(Of course I'm biased being on of the OpenSC developers
Every three years I look at all the latest shiny hardware and write down the nicest
machine I can think off, totaly bleeding edge, highest capacity and everything,
but still something I think that might work. Then I put that paper aside.
6 months later those components are widely available at a reasonable price, drivers
are stable and I can afford it, so I buy a new machine.
Having a new additional machine allows me to slowly move over to it, then take the
three old one once I no longer need it and reinstall it as server, so the 6 year old can slowly retire. Works fine for me, not much hardware trouble, and still reasonable costs.
already in dapper...
but that means usually a difficult choice: often there is an open source driver, but it is not fast.
On the other hand the super fast optimized proprietory driver is a maintenance nightmare: you need to
update it each time you update the kernel or Xorg and keep all those parts in sync.
Closed source binary drivers do not need to go away. But currently they do everything in the closed
binary only part. Please change that to only do those parts in binary only that are real important
(like all those secret tricks to get fast. And then open the rest - all that
boring code that switches graphic modes and checks what monitor is attached and turns of the backlight
if I close my laptop. There is no reason to keep those simple parts secret, and opening them would reduce
the driver to a much smaller size, most likely move the proprietory parts into userspace / libGL and Xorg.
That would be a good step, make maintenance of kernel and Xorg and friends easier, and still allow NVidia
and ATI to keep most of there secrets.
Come on ATI and NVidia, you can do it! Divide your graphic drivers into an kernel part and a user
space part and open source the kernel part. There are no big trade secrets in switich a graphic card
from mode A to mode B or finding out how fast it is clocked, what kind of monitor is attached and
all that.
If you absolutely want to keep a proprietory binary only core under a closed licensed, that is fine.
But please limit it to the real important stuff, like those mega tuned GL libraries and 3d accelleration,
and make the basic parts of your driver open source. That way kernel and big parts of X11 and libs
could be open source, and you could have additional, optional parts for X11 for better accelleration and
stuff like that.
Well, ATI is already a good step into that direction, the radeon driver works pretty well and is only
half as fast as the fglrx driver when it comes to video play back. Now please look at your code,
merge all the non important parts with radeon, and reduce the fglrx driver to some user space binary
blob. The kernel parts can't be that important, if radeon driver gets nearly everything right and fast.
nah. think about email encryption/decryption, that is far too heavy on the cpu,
your 386SX won't do it nicely. so lets move the encryption/decryption to a central
server.
oops. wait. aehm. never mind.
the crimimal part is not that bad - most prosecutors will simply drop small cases as not important,
so there is little change. and doing private copies for yourself is techincaly not legall, but
practical not enforceable.
the bad part is this: under the new law internet providers will have to answer to the media industry.
so far we had a real data privacy act, i.e. the policy / prosecutors had to ask the internet providers
for data, and then the media industry could have a look (and use it to start a civil law case).
too bad the justice department gave in and wants to save the prosecutors some work. as side effect
there is no way to stop the media industry from getting that data, there is no control, no one to
stop them from overusing that power etc. lets hope the supreme court will sack that part of the law
as conflicting with the german bill of rights (grundgesetz).
note: so far there is only a consensus within the governemnt and the lobby groups. german bundestag
does not have much power - most politicians simply do what their party tells them to do, as they have
next to no power without the party - but there is still hope they resist and successfully demand changes
before passing that proposed new law.
did you notice how the first half talks about movies and the fear of piracy,
but the second half only mentions the music industry not making as much money
as they used to?
well, the german movie industry has their own association with a web site at
http://www.bvv-medien.de/, and despide a very, very aggressive anti-consumer
anti piracy campain, they still more than doubled their revenue in the last
five years: 860 mio euro in 1999 vs. 1747 mio in 2004.
I guess noone of the german movie industry will read this, but: if I'm in cinema
and about some movie, and you want to show some ad to me, it should start with the
word "Danke" (thanks). After all I already paid for the movie ticket. Instead they
show some anti piracy ad with people send into jail and about to be raped or similar
stuff.
so far the ministry of justice has a proposal for a changed law.
that doesn't mean the bundestag will accept that proposal without changes.
at least joerg tauss of spd said there will be further long discussions,
and no simply voting on it. so lets hope for the best.
> February 31st is NOT a date
nor is Feburary 30th. But german banks charge interest for the 30th of
February. all the time. legal. even approved by the highest court.
(something about exponential mathematics being too hard for the average
citizen, so calculating with 360 days a year and linear interest is better).
You need to be more flexible if you want to target customers like german
banks as well.
I tried to - and these bugs annoy me:
* zip files: the build in stuff does not work, double clicking on a zip file creates a new window of
a process that hangs. you need to kill it using task manager. 7zip&co to the rescue.
* "security" implemented badly: everytime anything gets executed you get a prior popup where you need
to allow that. for example when I want to run the control panel, I get that popup first. or clicking
on anythign within the control panel. even worse are the details: they tell me that a dll is about
to be run as program. which dll? which application triggered that? of course that information is
missing. and of course, vista does not remember my choice. next time I run the same application, I
get the same popup. stupid.
* slow as hell, sometimes. also sometimes astonishing fast. a very different mix, it takes time to get
used to it. he slow parts of course are annoying, and the faster parts don't cancel out that effect.
* looks like mac os X. sure, some people like it, I prefer kde. and all the parts on mac os X look and
feel that I hate are now also implemented in vista. great.
* auto-run still active: autorun for cds was a bad idea. the sony/drm/rootkit story should have told them
at last. but they didn't learn it enough, so now you get a popup whether or not to run some autorun
(or some other software). kde has the same popups, they are a bad idea like almost every popup.
animeate some icon if you want or something, but don't move some software in the front.
* graphics don't work well - yet. the nvidia driver shipped with it doesn't work well, the result has
strange graphic effects that noone wanted (I tried only WoW, but I guess other apps will have them too).
nvidia has a driver for vista 32bit! but it doesn't install - finds no supported hardware. crap.
* sound problems? the driver works fine, the test tool works fine, but applications like WoW stay silent.
maybe the vista audio api broke and is no longer compatible?
* doesn't shut down. the start menu has a button that looks like it will shut down and turn of the
machine, and I was impressed how fast it is. but it does not do that, vista only suspends to disk.
* and of course for everyone (including me) to judge a product by a beta 2 version. oops, it is not beta2,
it is only community technlogy preview with buils 5270 - guess microsofts own QA department knows well
that while some parts have already the "beta 2" label, it could be called at best "alpha 2" if you take
the (missing) usability into account.