this seems to be a quite well written paper (as far as I can see from the Register's summary, the server is/.'ed).
Everything I read there points out things I don't like on windows, much better than I am capable of. While there exist many papers pointing out these things, they are often to "evangelistic" to be seriously considered for convincing management types.
I'm eager to get the whole document, it might have its worth even without mentioning the originaters (watch the copyright, though).
Why, in a week where we have alerts for Samba, php, kde (libs and network) and apache, do we have to hear about IE yet again?
Because samba et.al. use a completely different security philosophy. This shows and proves something that many people have said before, namely that MS' security philosophy based on "trust us, we know better what to do" is flawed. In the light of this news you can only laugh about popups like "Always trust content from microsoft corp.".
This is also not very encouraging for MS' auto-update feature in XP, and their whole fucking ideas of stuff in their OS's downloading components from the net without asking the user.
Note that the above is also true for other software publishers, but MS takes the spotlight for various reasons, like their omnipresence and their bullheadedness concerning these problems.
From my own experience my company uses FreeBSD. Yahoo uses FreeBSD. pair.com one of the bigger hosting providers uses FreeBSD. mp3.com uses FreeBSD. Hotmail used to use FreeBSD, I'm not sure if they still do. (after it was bought by MS that is). ftp.cdrom.com uses FreeBSD.
I don't want to argue about the merits of FreeBSD, because I don't much about it. But your argument is probably quite weak, because these companies decided to go with FreeBSD years ago. In the worst case (for FreeBSD) it just shows that at the time when they made their decision for FreeBSD it was superior, and today it's just "good enough" for not making the switch to linux. I'm not too convinced that FreeBSD _today_ is a better OS for servers which a lot of CPU power for application servers and such stuff. The scalability work done on linux 2.4/2.5 seems to imply that it would be a better choice for such workloads.
Please disregard the above lapsus... I hope that this is more than just a bid to recapture lost market share, but a real committment to play fair and adhere to open, published, and somewhat popular standards.
Well certainly it is just a coincidence that they do this now that they seriously feel the pain.
I hope that this is more than just a bid to recapture lost market share, but a real committment to play fair and adhere to open, published, and somewhat popular standards.
You confuse "not getting in linus' tree" with "not existant". No vendor kernel (exception is debian) is using the pure linus tree, and there is a plethora of projects offering custom (patched) kernels. Add to that the possibility to roll your own kernel with drivers you download from the vendors (programmers) website, which in most cases you can even build as a module (read: completely unintrusive), which takes away one advantage of microkernels.
I'm by no means an expert (esp. not for micro-kernels, that's why I ask), just a regular linux-kernel lurker, and reading all these arguments for micro-kernels leave me wondering if the theory isn't a bit behind the hardware development.
I see linux kernel developers seeking solutions for very complicated hardware architectures:
Hyperthreading
multi-core cpu architectures (IBM Power4)
"non-homogenous" memory architectures for multiprocessor machines (NUMA, etc.)
more and more "intelligent" storage devices
high-performance 3d graphics
soft/hard real-time performance
and so on..., where all of these things can occur combined.
I'm wondering if a micro-kernel is theoretically as good as a monolitic one in mastering the above difficulties - and if code for all of this had to be integrated in a microkernel, and if this thing really should be called microkernel after that. And as far as I know, QNX isn't really an example for an OS which shines on an 8-way box.
Another thing: You do take a performance hit, but in a world where Java, Perl, and XML are used for production work, it's tiny by comparison.
This I doubt, because there are quite a few things an os can do to help gain performance for typical application tasks. See for instance ingo molnars work on thread startup performance -100000 threads in a very short time (some secs) compared to some minutes on linux before that. This work was done because of (Oracle/Sun)'s needs for their application - and java needs everything fast which is related to threads.
Um, most users can also figure out how to download the whole thing too.
I assure you, most is not enough in most cases. People left klick, some movie player pops up, and they wonder why the movie is so slow on their 56k. And starcraft is not a good enough criterium to filter out enough "lusers".
As for not being on a streaming server, well, this won't help for 56k too, but you are right, I wouldn't want to keep this marketing pap on my HD too. Maybe they think that people will pass it around etc., and that streaming is more expensive, I don't know.
What I wanted to just point out is that zipping is not always used for compression only, nearly always thanks to braindead internet explorer.
See for yourself, but beware, it will make your brain hurt.
That's cause of Divx. On another point....WHY ARE THE FRIGGIN CLIPS ZIPPED! Zipping a Divx file gets you practically nothing. IN fact it probably adds to the filesize.
Yes I'm offtopic, but I'd blame that on internet explorer. Sometimes the best (easiest) solution for preventing it from opening the file in an external app (Movieplayer in this case) without asking the user is to zip the file. I guess they don't want one million people sending them angry mails because the "streaming video" is so slow.
Imagine, fighting against 8000 bots each controlled by 8 processors. Factoring in processor speed, that makes each bot at least 2 times more clever than the machine that recently gained a draw in chess against kramnik.
except that the 3rd party software here could send the MAC addresse of the cable modem on the local lan as easy as the MAC address of the computer it's installed on.
Posting this in the US would not be a violation of the DMCA except if you used some ludicrously tortured logic.
I'm not going to argue with you about Felton, Skylarov and stuff. But let's look at it from another angle:
RHAT is a public company, i.e. they have shareholders
Because of 1, they have a responsibility to do the best for the present/future of their company, i.e. not playing silly political games.
Doing what they did do can have a bad effect on their reputation (unamerican, non-patriotic, pro-terrorists - you get the drift), which in turn will negativly affect their ability to make profit
Not doing what they did do might have also bad consequences, i.e. core developers get arrested.
They have lawyers.
Combine 1 and 2 with 5, and you come to the conclusion that you have to ask lawyers in order to cover your ass.
The most useful statement of the principle for scientists is,
"when you have two competing theories which make exactly the same predictions, the one that is simpler is the better."
Might I add something to the last sentence? "...the one that is simpler is the better, unless the other one promises fundamentaly more funding and/or publicity"
I really can't understand all this whining from people who are in no way negativly impacted from the decision to use cvs. And as far as I have read the pro-bitkeeper people are going to great lengths to make life easy for developers not willing to use bitkeeper.
If the majority of linux kernel developers decided that they need to sacrifice a virgin every month, that would be fine with me - the virgin's consent provided. Well, perhaps if they called her GNU/virgin...
Better reread what I quoted and what the parent poster wrote. With the exception that yast is indeed not GPL, everything else you said is factually wrong.
You may redistribule the files on the source CD, so long as you only redistribute the non-commercial parts. Redistributing Yast will get you in deep legal trouble.
Wrong. FUD. read the fucking license terms. I don't know why people feel obliged to spout off such a nonsense. I shouldn't do it, but anyway, I digged out something which your eyes clearly have never seen before, the yast license:
It is forbidden to reproduce or distribute data carriers which have been reproduced without authorisation for payment without the prior written consent of SuSE Linux AG or SuSE Linux. Distribution of the YaST programme, its sources, whether amended or unamended in full or in part thereof, and the works derived thereof for a charge require the prior written consent of SuSE Linux AG.
So check your facts next time, please!
Re:Recycle Bins - don't you just hate them?
on
Undelete In Linux
·
· Score: 2
Show your friend how much nicer it were if he could find out if something needs a certain file by knowing when had been accessed in the past, instead of moving it and waiting for random breakage.
Because like it or not, Vignette, Cumulus, Mediasphere, and QPS and so forth, whatever they may cost to buy, will all continue to be cheaper in the commercial world if the open source alternatives require coding to do jobs that commercial products better enable with the click of a button.
Hmm, I like what you said, but here I'm sceptical. Anyone I've talked to who did research into commercial CMS - as deep as taking courses held by the company selling the CMS - was complaining about the fact that these systems, while doing some stuff out of the box with a click of a button, always require quite a bit of customization (i.e. programming) to solve the problems the clients really have.
But you're right that commercial products are better in giving the impression of solving all problems out-of-the-box.
Slashdot Mirror
Hmm, this explanation doesn't fit well with what I read at the Reg:
The whitepaper, by MS Windows 2000 Server Product Group member David Brooks,
Whereas in Win2K: "Some parameters that control the system's [...]
Cleary, the original hotmail guys wouldn't have thought about W2k, which was non-existant at that time.
The team was unable to reduce the size of the image below 900MB
Dito, I doubt any MS operating system's image at that time couldn't be reduced to less than 900MB.
They also mention Advanced Server, that "at" is deprecated, Interix 2.2 and so on.
No, I doubt your are right.
this seems to be a quite well written paper (as far as I can see from the Register's summary, the server is /.'ed).
Everything I read there points out things I don't like on windows, much better than I am capable of. While there exist many papers pointing out these things, they are often to "evangelistic" to be seriously considered for convincing management types.
I'm eager to get the whole document, it might have its worth even without mentioning the originaters (watch the copyright, though).
This begs the question why they did implement this trust "feature" in the first place.
Why, in a week where we have alerts for Samba, php, kde (libs and network) and apache, do we have to hear about IE yet again?
Because samba et.al. use a completely different security philosophy. This shows and proves something that many people have said before, namely that MS' security philosophy based on "trust us, we know better what to do" is flawed. In the light of this news you can only laugh about popups like "Always trust content from microsoft corp.".
This is also not very encouraging for MS' auto-update feature in XP, and their whole fucking ideas of stuff in their OS's downloading components from the net without asking the user.
Note that the above is also true for other software publishers, but MS takes the spotlight for various reasons, like their omnipresence and their bullheadedness concerning these problems.
From my own experience my company uses FreeBSD. Yahoo uses FreeBSD. pair.com one of the bigger hosting providers uses FreeBSD. mp3.com uses FreeBSD. Hotmail used to use FreeBSD, I'm not sure if they still do. (after it was bought by MS that is). ftp.cdrom.com uses FreeBSD.
I don't want to argue about the merits of FreeBSD, because I don't much about it.
But your argument is probably quite weak, because these companies decided to go with FreeBSD years ago.
In the worst case (for FreeBSD) it just shows that at the time when they made their decision for FreeBSD it was superior, and today it's just "good enough" for not making the switch to linux.
I'm not too convinced that FreeBSD _today_ is a better OS for servers which a lot of CPU power for application servers and such stuff. The scalability work done on linux 2.4/2.5 seems to imply that it would be a better choice for such workloads.
Maybe because it's the greatest non-trivial 8 bit palindrom?
...
Probably not
...getting a fixed IP costs way more than a .mac account.
Well, I can sell you 192.168.0.1 really cheap, just mail me you kredit card number.
Livetime access granted.
Please disregard the above lapsus ...
I hope that this is more than just a bid to recapture lost market share, but a real committment to play fair and adhere to open, published, and somewhat popular standards.
Well certainly it is just a coincidence that they do this now that they seriously feel the pain.
I hope that this is more than just a bid to recapture lost market share, but a real committment to play fair and adhere to open, published, and somewhat popular standards.
Thanks for answering the questions.
You confuse "not getting in linus' tree" with "not existant".
No vendor kernel (exception is debian) is using the pure linus tree, and there is a plethora of projects offering custom (patched) kernels. Add to that the possibility to roll your own kernel with drivers you download from the vendors (programmers) website, which in most cases you can even build as a module (read: completely unintrusive), which takes away one advantage of microkernels.
I see linux kernel developers seeking solutions for very complicated hardware architectures:
and so on..., where all of these things can occur combined.
I'm wondering if a micro-kernel is theoretically as good as a monolitic one in mastering the above difficulties - and if code for all of this had to be integrated in a microkernel, and if this thing really should be called microkernel after that.
And as far as I know, QNX isn't really an example for an OS which shines on an 8-way box.
Another thing:
You do take a performance hit, but in a world where Java, Perl, and XML are used for production work, it's tiny by comparison.
This I doubt, because there are quite a few things an os can do to help gain performance for typical application tasks. See for instance ingo molnars work on thread startup performance -100000 threads in a very short time (some secs) compared to some minutes on linux before that.
This work was done because of (Oracle/Sun)'s needs for their application - and java needs everything fast which is related to threads.
Um, most users can also figure out how to download the whole thing too.
I assure you, most is not enough in most cases. People left klick, some movie player pops up, and they wonder why the movie is so slow on their 56k. And starcraft is not a good enough criterium to filter out enough "lusers".
As for not being on a streaming server, well, this won't help for 56k too, but you are right, I wouldn't want to keep this marketing pap on my HD too. Maybe they think that people will pass it around etc., and that streaming is more expensive, I don't know.
What I wanted to just point out is that zipping is not always used for compression only, nearly always thanks to braindead internet explorer.
See for yourself, but beware, it will make your brain hurt.
I
That's cause of Divx. On another point....WHY ARE THE FRIGGIN CLIPS ZIPPED! Zipping a Divx file gets you practically nothing. IN fact it probably adds to the filesize.
Yes I'm offtopic, but I'd blame that on internet explorer. Sometimes the best (easiest) solution for preventing it from opening the file in an external app (Movieplayer in this case) without asking the user is to zip the file. I guess they don't want one million people sending them angry mails because the "streaming video" is so slow.
Someone from IBM recently offered they host _and_ maintain a bugzilla database for the linux kernel.
That is a great service to offer.
Imagine, fighting against 8000 bots each controlled by 8 processors.
Factoring in processor speed, that makes each bot at least 2 times more clever than the machine that recently gained a draw in chess against kramnik.
Wow
except that the 3rd party software here could send the MAC addresse of the cable modem on the local lan as easy as the MAC address of the computer it's installed on.
I'm not going to argue with you about Felton, Skylarov and stuff. But let's look at it from another angle:
Combine 1 and 2 with 5, and you come to the conclusion that you have to ask lawyers in order to cover your ass.
Now I have to ask: Are you are lawyer?
To quote your link:
The most useful statement of the principle for scientists is,
"when you have two competing theories which make exactly the same predictions, the one that is simpler is the better."
Might I add something to the last sentence?
"...the one that is simpler is the better, unless the other one promises fundamentaly more funding and/or publicity"
That should explain it...
ehm,
s/cvs/bitkeeper/
Damn right.
...
I really can't understand all this whining from people who are in no way negativly impacted from the decision to use cvs. And as far as I have read the pro-bitkeeper people are going to great lengths to make life easy for developers not willing to use bitkeeper.
If the majority of linux kernel developers decided that they need to sacrifice a virgin every month, that would be fine with me - the virgin's consent provided.
Well, perhaps if they called her GNU/virgin
Better reread what I quoted and what the parent poster wrote.
With the exception that yast is indeed not GPL, everything else you said is factually wrong.
Wrong. FUD. read the fucking license terms.
I don't know why people feel obliged to spout off such a nonsense.
I shouldn't do it, but anyway, I digged out something which your eyes clearly have never seen before, the yast license:
So check your facts next time, please!
Show your friend how much nicer it were if he could find out if something needs a certain file by knowing when had been accessed in the past, instead of moving it and waiting for random breakage.
Then teach your friend what
ls -lu
means.
Because like it or not, Vignette, Cumulus, Mediasphere, and QPS and so forth, whatever they may cost to buy, will all continue to be cheaper in the commercial world if the open source alternatives require coding to do jobs that commercial products better enable with the click of a button.
Hmm, I like what you said, but here I'm sceptical. Anyone I've talked to who did research into commercial CMS - as deep as taking courses held by the company selling the CMS - was complaining about the fact that these systems, while doing some stuff out of the box with a click of a button, always require quite a bit of customization (i.e. programming) to solve the problems the clients really have.
But you're right that commercial products are better in giving the impression of solving all problems out-of-the-box.