Why Linux and open source sucks

[Mike]

Open source programs in general suck pretty bad also. Here are some
loose facts regarding why projects like Linux, PHP, MYSQL, and
other open source programs suck:

- Linux is simply a clone of an operating system over 20 years
old similar to DOS (Unix). That's ALL Linux is. The kernal was taken,
decompiled, and stolen by Linus himself. That's a fact. Xerox
should have sued, but at the during the early 1991, no one was
really using Unix and Xerox had better things to do with their
recources than to sue someone steeling something what Xerox
doesnt care about. (ie: You can download lots of old games online
that were sold in stores, but these companies don't
care cause these games were in the past and don't currently make
them money so they really don't care what happens to them).

- Other open source project are MySQL, PHP, Perl, Apache, Mozilla..
These are key components of business such as web serving and database.
When Microsoft was smaller (around mid 90's), these products killed
Microsoft in performance and usablility. They were
the top choice for services and MS had crappier versions of the
same thing which no one liked or used.

- Eventually, Microsoft grew in money and power. They came out
with new OS's like Windows 2000 and then things began to change.
MS's SQL Server 2000 started beating out Oracle, IBM DB2,
and MySQL is speed and performance. People started to notice
this and began to migrate over to SQL Server 2000. Remeber that
business and the world are based off these 4 databases.

- iis 5.0 and ASP was realeased in 2000 and once again, the
benchmarks smoked competition. They *killed* Java, and beat out
PHP. An independant magazine developed identicle e-Commerce
web sites (like this) in PHP, asp, and JSP.

JSP - 17 pages per second.
PHP - 54 pages per second
asp - 80 pages per second (and with acceleated code and compression
was up to 160 pages)

Companies around that time popped up to push for Linux. One was
Red Hat Linux. This was a serious effort to make an OS to compete
with Microsoft Windows. Poeple poured millions into Red Hat stock.
Cause at that time, ALL high tech companies where hot.

Remember, Red Hat is GNU. Thus, they have to give you their
source code on each product given away or sold. In reality, their
products were being sold at future shop for profit and i assure you,
their lawyears would be coming after you if you compiled their code
and sold it for profit.

Fact is, Red Hat turned out to be a very poor, slow OS. It had many
features, but they simply didn't work (ie: their plug and play).
Star Office emerged and turned out to be a very buggy, poor,
and *extremely* slow Office suite.

By 2000+ and XP, Microsoft IIS 6.0 and .NET, Microsoft is now
killing *any* open source project in performance, reliability, and
ease of use. Oracle recently announced that you may NOT
publish benchmarks if you use their product. (that's because they
know they are now the slowest database).

Recently at a yearly hacking/sercurtiy confrence, contestants entered
and people over the week would try to hack into the
servers. One was Microsoft running IIS 6.0. There were over
80,000 hack attempts and not one person could get in. They
won. Linux based systems were hacked into instantly.

However, Red Hat. BSD, Linux does have Apache. And they also
have one very, very, important thing going for them to make them a
good choice for dB and web services: -> They are NOT Microsoft.

??? you ask?

Well, once you get to speak to network admins all day like I do,
there is a *good* reason to go the BSD, Linux route. There is
one thing that keeps these poorer, slower OS's and services alive
and used in the industry:

Viruses don't work on them.

 

[Morten Wennevik]

And this would be relevant to this group in what way?

 

[Maziar Aflatoun]

Seek help... for production environment Linux/FreeBSD still rule...

 

[Mr.Tickle]

production ? Linux ; ROFL, we use QNX

Seek help... for production environment Linux/FreeBSD still rule...

 

[Paul E Collins]

[... snipped lots of irrelevant advocacy ...]
Oracle recently announced that you may NOT
publish benchmarks if you use their product.
(that's because they know they are now the
slowest database).

Then why did Microsoft prohibit the publishing of benchmarks involving SQL
Server?

P.

 

[batman1990]

Wow, there are just so many things here that are bullshit.

If MS is so great, why does Java own the app server market.

..NET is getting good, but come of it.

Also XEROX had nothing to do with UNIX, it came of of AT&T. For that
matter Windwos NT can be said to be a cousin of 30 year old VMS with a
MACH kernel to boot.

The legacy arguemnet is just so weak.

[... snipped lots of irrelevant advocacy ...]
Oracle recently announced that you may NOT
publish benchmarks if you use their product.
(that's because they know they are now the
slowest database).

Then why did Microsoft prohibit the publishing of benchmarks involving SQL
Server?

P.

 

[anonymous coward]

He doesn't know what he is talking about. Just search his email address in
Google Groups, and you'll see. A VB 'programmer'.

Wow, there are just so many things here that are bullshit.

If MS is so great, why does Java own the app server market.

.NET is getting good, but come of it.

Also XEROX had nothing to do with UNIX, it came of of AT&T. For that
matter Windwos NT can be said to be a cousin of 30 year old VMS with a
MACH kernel to boot.

The legacy arguemnet is just so weak.

[... snipped lots of irrelevant advocacy ...]
Oracle recently announced that you may NOT
publish benchmarks if you use their product.
(that's because they know they are now the
slowest database).

Then why did Microsoft prohibit the publishing of benchmarks involving SQL
Server?

P.

 

[_]

Windows and Linux are both obsolete:

http://abderaware.typepad.com/abderaware/2003/10/the_art_of_war.html




--
What the caterpillar calls the end,
the rest of the world calls a butterfly.
~ Lao-tzu

Components For Thinkers
www.abderaware.com
zane @at@ abderaware .dot. com

 

[Daniel O'Connell]

Windows and Linux are both obsolete:

http://abderaware.typepad.com/abderaware/2003/10/the_art_of_war.html

Quite true, however I suspect we're stuck with it for a while yet to come.
Copying an old design is easy(which both Linux and Windows did to an
extent), coming up with something wholly new is something else...

The problem may not be that its hard to come up with ideas, but that its
hard to come up with ideas other people like.

 

[_]

Daniel,

... coming up with something wholly new is something else...

There is no shortage of relevant research and various implementations to play
with.

The problem may not be that its hard to come up with ideas, but that its
hard to come up with ideas other people like.

But the situation on the desktop is even more serious. Even assuming a good
idea for a componentized, secure (http://blog.monstuff.com/archives/000102.html)
operating system, overcoming the stranglehold MS has on the desktop is The Major
Problem - IMO.

But let's try this idea: Suppose that you have an incrementally downloading OO
framework (perhaps derived from mono http://www.go-mono.com) and that the common
programs people currently pay MS a pile-o-cash for (word etc) could be replaced
with programs that run on that framework. And lots of shareware programs.

_Eventually_ wouldn't the OS become irrelevant?

The problem Linux, Lindows, BeOS, et al have had is that it's impractical to
knock Windows off the desktop overnight. A new strategy seems to be in order.



--
What the caterpillar calls the end,
the rest of the world calls a butterfly.
~ Lao-tzu

Components For Thinkers
www.abderaware.com
zane @at@ abderaware .dot. com

 

[Daniel O'Connell]

Daniel,


There is no shortage of relevant research and various implementations to play
with.

No, but I've yet to see anything I would deem as wholly new. Most research
projects I've read about(I won't claim to have read all of 'em, however, I
am generalizing here), while different than existing systems, are obvious
and have yet to prove to me that the idea is really better, from a usability
viewpoint.

But the situation on the desktop is even more serious. Even assuming a good
idea for a componentized, secure (http://blog.monstuff.com/archives/000102.html)
operating system, overcoming the stranglehold MS has on the desktop is The Major
Problem - IMO.

Perhaps, another problem is most researchers and hobbists don't even try to
create something that would have mass market appeal. Most people I know
wouldn't deal with it well at all. That is why MS has its stranglehold,
people understand Windows well enough to use it. It is far, far, far more
likely that these research projects will just work their way into a large
company(read: Microsoft) and eventually provide them with the tools to
create something akin to the research. But that will not be in short order.

But let's try this idea: Suppose that you have an incrementally downloading OO
framework (perhaps derived from mono http://www.go-mono.com) and that the common
programs people currently pay MS a pile-o-cash for (word etc) could be replaced
with programs that run on that framework. And lots of shareware programs.

_Eventually_ wouldn't the OS become irrelevant?

Wasn't that the goal of Java, in a sense?
And, incrementally downloading sounds nice, but it forgets that the majority
of computers are either not online or don't have a connection capable of
downloading components at any given time. No one would be happy if they had
to download a 10 meg component set just to run notepad(or a clone there of,
actually).
When it comes down to it, an OS itself is little more than a kernel and
associated services(drivers, etc), each currently existing OS provides
different kernel calls, different libraries, and capabilities. Windows, by
default, provides drawing routines with the system, that is a minor
difference but nothing of significant issue. The various apps that most
people consider Windows (Explorer.exe most specifically) can be changed,
fairly easily in most cases, to another application that provides the same
function. I do not consider explorer, IE, or any other such things as part
of the OS per se, but just as applications bundled with it.
There is really no need to change the underlying OS to make the OS
irrelevant, simply a framework that operates over all OS's(like Java).
However, that really didn't work, it limits the user a great deal, it sticks
capability to the lowest common denominator, and eventually starts to bind
the hands of innovation. Java(and .NET eventually) can change safely(due to
runtime side-by-side, etc), an Operating System framework of similar type
could never make a breaking change, or risk failing every application
written for it. That would result in either requiring a side-by-side OS
solution(which, in the end, would result in a kernel and a framework on top
of it, just as Java and .NET are today) or require leaving in broken,
useless, etc functionality(Windows STILL has Win16 functions available, just
shouldn't use them). It just doesn't solve any problems.

The problem Linux, Lindows, BeOS, et al have had is that it's impractical to
knock Windows off the desktop overnight. A new strategy seems to be in order.

The problem with Linux, Lindows, and BeOS is that they didn't provide
anything significantly(or at all) new. They, in many ways, used the same
metaphors, the same design concepts, the same desktop theory, the same basic
architecture(they are different, but not in any meaningful way), but they
did it in, for a user experianced with Windows, a harder way. Why would
anyone bother?
When it comes down to it, the kernel matters very very little these days, it
should be designed to do what it must and not require user applications to
be aware of its implementation, or really even its existance. Application
interaction, user interface, and system security are quite a bit more
important, and no mainstream operating system yet has taken any significant
strides towards providing anything new in those areas, its always the same
old thing, part deux.

There are alot of possiblities, and I have my own dreams(mono has been in my
mind of late as well...for one particular idea), but the question is will
anyone ever implement a new idea in a way that will actually matter to
anyone but the minority research community. A neat kernel with a commandline
is useless to 99% of the world, and perhaps borderline so to most of that
other 1%. Even if its taken up by the vast community, the system would take
a long time to mature. Look at linux...it is not old but not very new
either, 10 years may be needed to mature and learn a system desing well
enough to become mainstream.
It will take a long time. Microsoft, and Linux, will one day fall and be
looked back on as history. The question is still up in the air as to if
either or both will be a "What were we thinking..." thing, but that really
is a seperate issue.

 

[_]

Daniel,

... That is why MS has its stranglehold [because] people understand Windows
well enough to use it.

True, but for most people Windows is just another pretty face. Windows has
the desktop market because it was first there with a decent price on relatively
inexpensive hardware (yeah, I thought my first color pc-compatible with a 10mb
drive was inexpensive at $5000!).

Now that MS has a monopoly on the desktop prying it loose with anything, no
matter how familiar the UI, or inexpensive, seems to be nearly impossible.

Wasn't that the goal of Java, in a sense?

Good point. But as a framework upon which to run applications I think .NET is
better.

And, incrementally downloading sounds nice, but it forgets that the majority
of computers are either not online or don't have a connection capable of
downloading components at any given time. No one would be happy if they had
to download a 10 meg component set just to run notepad(or a clone there of,
actually).

A 10mb component or application is completely out of the question. But that's
not really what I'm thinking of, which is:

Take mono, for example, and rearrange such that it consists of more and smaller
assemblies. Define the core required by any mono app. And install _that_ on
demand with the first downloaded application; which also downloads any extras it
needs, such as a datagrid component or whatever.

When the next app is installed it won't require the core download, although it
may require other not-yet-downloaded assemblies.

One of the problems with the .net platform is that it's a monolithic install,
and everyone gets Web Services etc even if there is no way they will ever be
using them.

So, what I'm thinking of is much finer granularity downloaded by applications
installation programs.

When it comes down to it, an OS itself is little more than a kernel and
associated services(drivers, etc), each currently existing OS provides
different kernel calls, different libraries, and capabilities.

Or if you really want to have fun you can, essentially, get rid of the OS:

http://www.pdos.lcs.mit.edu/exo.html

The various apps that most
people consider Windows (Explorer.exe most specifically) can be changed,
fairly easily in most cases, to another application that provides the same
function. I do not consider explorer, IE, or any other such things as part
of the OS per se, but just as applications bundled with it.

Too right. And that's the core of the argument that MS is abusing its monopoly
power.

There is really no need to change the underlying OS to make the OS
irrelevant, simply a framework that operates over all OS's(like Java).
However, that really didn't work, it limits the user a great deal, it sticks
capability to the lowest common denominator ...

I don't think that's necessarily the case. If the framework replaces the OS, at
least in an abstract sense, then it is the common demoninator. So it binds the
hands of the innovators no more than the OS does, and considerably less IMO
since new, finely-grained, components can be downloaded at any time.

The problem with Linux, Lindows, and BeOS is that they didn't provide
anything significantly(or at all) new.
Right.

There are alot of possiblities, and I have my own dreams(mono has been in my
mind of late as well...for one particular idea), but the question is will
anyone ever implement a new idea in a way that will actually matter to
anyone but the minority research community.

The one hope I see is this: If a framework makes it easy for shareware and
other small-scale developers (keeping my self-interest in mind here <g>) can
easily distribute highly functional applications developed with the latest in
development tools, but without requiring a 20mb download, then they could drive
the adoption of a framework under which a truely secure component-based
operating system could be installed - for those who care about such things.

It will take a long time. Microsoft, and Linux, will one day fall and be
looked back on as history.

I'll kneel down behind them, you push!


--
What the caterpillar calls the end,
the rest of the world calls a butterfly.
~ Lao-tzu

Components For Thinkers
www.abderaware.com
zane @at@ abderaware .dot. com

 

[Daniel O'Connell]

Daniel,

... That is why MS has its stranglehold [because] people understand Windows
well enough to use it.

True, but for most people Windows is just another pretty face. Windows has
the desktop market because it was first there with a decent price on relatively
inexpensive hardware (yeah, I thought my first color pc-compatible with a 10mb
drive was inexpensive at $5000!).

Now that MS has a monopoly on the desktop prying it loose with anything, no
matter how familiar the UI, or inexpensive, seems to be nearly impossible.

Yes, it does, but MS is still providing innovation(although alot of people
can rarely grasp that) and may actually end up providing the soultions we
are all looking for.

Good point. But as a framework upon which to run applications I think ..NET is
better.

So do I, but its not yet quite designed to operate across all systems, so it
doesn't fit.

A 10mb component or application is completely out of the question. But that's
not really what I'm thinking of, which is:

Take mono, for example, and rearrange such that it consists of more and smaller
assemblies. Define the core required by any mono app. And install _that_ on
demand with the first downloaded application; which also downloads any extras it
needs, such as a datagrid component or whatever.

When the next app is installed it won't require the core download, although it
may require other not-yet-downloaded assemblies.

One of the problems with the .net platform is that it's a monolithic install,
and everyone gets Web Services etc even if there is no way they will ever be
using them.

So, what I'm thinking of is much finer granularity downloaded by applications
installation programs.

That I would like to see and I hope the .NET framework itself provides that
level of support in the near future. Most of hte time I only need
System.dll, mscorlib, System.Xml, System.Data, and System.Windows.Forms. I
would like to see a 5-8 meg runtim with downloadable components, via a nice
UI and via a nice installer system. I'd also like to see a framework that
doesn't require as complex an install as the current one. But that remains
to be seen...

Or if you really want to have fun you can, essentially, get rid of the OS:

http://www.pdos.lcs.mit.edu/exo.html

I have considered this myself, I dunno how realistic it is, as it provides
some strange development isuses and greatly expands the surface inwhich a
serious bug could exist.

Too right. And that's the core of the argument that MS is abusing its monopoly
power.

To an extent, it is no where near as abusive as, say, apple, Apple just
doesn't have the mrket share for anyone to care. IE, explorer, etc can be
replaces with other shells, browsers, etc. I think the issue is more that
its not possible to easily change the default by OEM's(Although, even if an
OEM could, by doing so they would probably kill off a large portion of their
user base).

I don't think that's necessarily the case. If the framework replaces the OS, at
least in an abstract sense, then it is the common demoninator. So it binds the
hands of the innovators no more than the OS does, and considerably less IMO
since new, finely-grained, components can be downloaded at any time.

Consider the problem with java, alot couldn't be done because platform
issues weren't even. Native interop was annoying, the windowing system
sucked, etc. Placing the framework AS the OS would simply result in limiting
the system to the capacities of the framework, and applications that wised
to target a wide range would have to target the lowest common denominator of
all the framework implementations(which, I suspect, would be many in time).

The one hope I see is this: If a framework makes it easy for shareware and
other small-scale developers (keeping my self-interest in mind here <g>) can
easily distribute highly functional applications developed with the latest in
development tools, but without requiring a 20mb download, then they could drive
the adoption of a framework under which a truely secure component-based
operating system could be installed - for those who care about such things.

Yes, a compontized operating system is desireable, no system yet really
provides it. That is something close to what I am thinking about. It seems
that alot of functionality should exist. Standard interfaces for word
processors, email apps, text editors(which word processor should derive
from), even calculators. Standard methods to control volume, etc. But I
worry far more about application interaction than most other things. I also
do not think that being component based will nessecerly create security or
result in any better security than non-componentized system would. There are
still a great many possible issues and buffer overflows are to common across
the board.

 

[_]

Daniel,

Yes, it does, but MS is still providing innovation(although alot of people
can rarely grasp that) and may actually end up providing the soultions we
are all looking for.

MS does innovate, but not in ways that would undermine its control of the
desktop. No surprise there really, it is the nature of large corporations to
try to perpetuate themselves.

That I would like to see and I hope the .NET framework itself provides that
level of support in the near future.

I wouldn't count on it. If MS did that then it wouldn't be so able to push OS
upgrades down the throats of businesses. For example: I recently spoke with a
friend doing consulting for some insurance company. They require all of their
agents to upgrade to XP so they can use the .NET software being written. MS
scores again.

Most of hte time I only need
System.dll, mscorlib, System.Xml, System.Data, and System.Windows.Forms. I
would like to see a 5-8 meg runtim with downloadable components, via a nice
UI and via a nice installer system.

Exactly. Although my estimation is that it would be more like 4-5mbs. Crude
estimate, and wrong by some unknown percentage.

I'd also like to see a framework that
doesn't require as complex an install as the current one. But that remains
to be seen...

Why would MS do that? If they really cared about getting the OS on W98 and up
then they would either have a light version, or be getting CDs into peoples'
snailmail boxes.

I have considered this myself, I dunno how realistic it is, as it provides
some strange development isuses and greatly expands the surface inwhich a
serious bug could exist.

There are lots of possibilities, EROS for example:

http://www.eros-os.org/essays/00Essays.html

That addresses some of the security issues you raised later in the post I reply
to.

IE, explorer, etc can be
replaces with other shells, browsers, etc.

Sure, they can. But how many end-users are going to go through the song and
dance when the MS apps work well-enough?

Consider the problem with java, alot couldn't be done because platform
issues weren't even.

This problem is as old as ascii-terminal gui systems, probably older. But
ascii-terminals really did have widely varying capabilities. PC hardware is a
lot more similar from one machine to the next, with appropriate abstractions and
drivers as even an obsolte monolith such as windows demonstrates!

Native interop was annoying, the windowing system
sucked, etc. Placing the framework AS the OS would simply result in limiting
the system to the capacities of the framework ...

I wonder how true, as a generality, that necessarily is. The places I've used
interop heavily are those places where MS cut-corners in the framework. Take
their Socket class for example. It provides no ssl support - despite the fact
that decompiling their code shows that the intent was there to provide it, some
day. Adding it to your own code requires heavy use of interop ... or it requires
an openssl in managed code.

But when you get right down to it, the ssl issue high-lights how fuzzy the
definition of "operating system" is. I've used openssl and to me it's just
another application library which could just as well be a set of components.
Once upon a time operating systems were concerned with allocating and providing
access to hardware resources. Windows and the other monolithic operating
systems have bitten off some huge chunks of turf which don't necessarily belong
to them.

But I worry far more about application interaction than most other things.
I also do not think that being component based will nessecerly create security or
result in any better security than non-componentized system would. There are
still a great many possible issues and buffer overflows are to common across
the board.

What do you think about: http://blog.monstuff.com/archives/000102.html ?



--
What the caterpillar calls the end,
the rest of the world calls a butterfly.
~ Lao-tzu

Components For Thinkers
www.abderaware.com
zane @at@ abderaware .dot. com

 

[Daniel O'Connell]

Daniel,


MS does innovate, but not in ways that would undermine its control of the
desktop. No surprise there really, it is the nature of large corporations to
try to perpetuate themselves.

That is actually the nature of all entities; any company is out to
perpetuate itself, otherwise why bother?

I wouldn't count on it. If MS did that then it wouldn't be so able to push OS
upgrades down the throats of businesses. For example: I recently spoke with a
friend doing consulting for some insurance company. They require all of their
agents to upgrade to XP so they can use the .NET software being written. MS
scores again.

Honestly, if they were using Win98\ME, then they needed to upgrade anyway,
just from a security perspective, .NET works fine on 2k as far as I
know(everything I've ever writen has worked just fine anyway). There is a
logic behind not bothering with the Win9x series of systems, it makes
support and testing far more expensive for little benifit(and it actually
probably helps bind us developers to a lower common denominator, once
again).

Exactly. Although my estimation is that it would be more like 4-5mbs. Crude
estimate, and wrong by some unknown percentage.

I didn't check, 5-8 just came to mind.

Why would MS do that? If they really cared about getting the OS on W98 and up
then they would either have a light version, or be getting CDs into peoples'
snailmail boxes.

Well, currently, the installation has alot of baggage, I don't know if MS
has any reason to change it, Mono certainly followed in the footsteps
though(its not exactly xcopy deployable either, I use an MSI to install it).
Personally, I don't worry about Win98 anyway. The number of classes, etc I
have to give up on to support win98 is far to restrictive to make me even
care. But then I'm not a shareware author.

There are lots of possibilities, EROS for example:

http://www.eros-os.org/essays/00Essays.html

That addresses some of the security issues you raised later in the post I reply

Sure, they can. But how many end-users are going to go through the song and
dance when the MS apps work well-enough?

Thats up to the end user, I seriously doubt that common users using Redhat
Linux would jump up to change shells either, beyond what the installation
gives them. I know people still using fwwm95 simply because its the default
on their system(and they are using aincient versions of linux, at that). The
default will always be hte most popular...how would you like windows with no
web browser(how would you get a web browser? ftp.exe and guessing about a
possible server?) and no shell? Or if you bought a Dell you end up with
DellXplore that works kinda like Explorer, but on your friends gateway you
have to use something far more like MacOSX's shell? That kind of crap really
doesn't make me want to use a platform nor does it promote system
integration, instead you end up with thousands of seperate parts that will
never work together because there is no common ground. Such a situation is a
degredation of use, it makes support massivly difficult, and would lock a
company into a specific vendor(it would likewise be a pain to have 50%
Dell's and 50% Gateways, with different UI's). OEM customization really
should never go beyond some labels, or they begin to complicate the system
needlessly.

This problem is as old as ascii-terminal gui systems, probably older. But
ascii-terminals really did have widely varying capabilities. PC hardware is a
lot more similar from one machine to the next, with appropriate abstractions and
drivers as even an obsolte monolith such as windows demonstrates!

Hardware isn't the issue, its what the system supports. Infact hardware
should be utterly ignored by this point, aside from disk space, monitor
size(in some cases), and printer properties. If you are willing to write
every capability you want from scratch, then there is no issue. If you
aren't, then you are locked to what you are willing to write, the libraries
you can get, and what the system provides. Some stuff can only be provided
by the system(and components plugged into the system). One major problem is
System A doens't support extensions in the same manner as System B and you
have to use two different extensions and two different code bases to provide
the support across the two systems...thats not fun. Standardized interfaces
means ALOT more than the hardware. Just look at a standard unix configure
script, interfaces are vastly unstandardized along libc and the like, and it
makes life annoying, even if tools exist to test for it. We escape that in
windows only because of one providing vendor, how long is it going to be
before mono & the MS framework are no longer in sync and you have to check
what you are running under and change code paths based on what will work?
That is what I worry about.

I wonder how true, as a generality, that necessarily is. The places I've used
interop heavily are those places where MS cut-corners in the framework. Take
their Socket class for example. It provides no ssl support - despite the fact
that decompiling their code shows that the intent was there to provide it, some
day. Adding it to your own code requires heavy use of interop ... or it requires
an openssl in managed code.

But when you get right down to it, the ssl issue high-lights how fuzzy the
definition of "operating system" is. I've used openssl and to me it's just
another application library which could just as well be a set of components.
Once upon a time operating systems were concerned with allocating and providing
access to hardware resources. Windows and the other monolithic operating
systems have bitten off some huge chunks of turf which don't necessarily belong
to them.

On this I agree as a generality, componentazation is a nessecity, OpenSSL
itself is a library, as far as i'm concerned. I wish the framework did
provide SSL support, I've heard it suggested that a SSL Client class may
exist in a coming framework version, but nothing about socket level SSL
communication. That is really an issue of design, I'd argue that SSL should
exist as a TcpClient deriv than as a special socket(a socket is a socket,
SSL, as far as I'm concerned, is layerd ontop of that).
However, that isn't the point. The smaller the framework the less an app can
do and assume it will run across all frameworks, it would be really naive to
assume that only one implementation would exist, no matter how light you
make it, someone else will want it lighter or think they could do it better.

However, I do think about a system where the kernel is just another
component that provides X facilities and a set of well defined interfaces
for extensions. A standard that worked in that way allowing platform
neutral, abstracted drivers, etc would be a wonder. No current system can
handle that, every one requires either kernel or kernel mode hacking or
other complicated, semi-well defined C style interfaces. In such a system,
like you've alluded too, nothing should ever execute in kernel mode, imho,
other than the very nessecities. Thread switching, device access, etc.
Filesystems and other such higher level stuff should exist as kernel
components in user space. However, that too may have significant performance
realities that I'm not aware of.

What do you think about: http://blog.monstuff.com/archives/000102.html ?

I think it shows a possiblity, but I also do not think that the system would
have to be componentized to use a capability design. Nor do I think that
simply because a system users a capability security arch. that it is
inherently secure. Mistakes can still exist, manners in which capabilities
can be incorrectly granted, compiler errors which do allow buffer overflows,
etc etc.

 

[Daniel O'Connell]

I think it shows a possiblity, but I also do not think that the system would
have to be componentized to use a capability design. Nor do I think that
simply because a system users a capability security arch. that it is
inherently secure. Mistakes can still exist, manners in which capabilities
can be incorrectly granted, compiler errors which do allow buffer overflows,
etc etc.

I also meant to point out that it would still be possible for an error to
exist, even in a well contained system, where an object that does have a
capability could be compromised. It may not be as easy, but that doesn't
mean its safer.

 

[Marc Scheuner [MVP ADSI]]

- Linux is simply a clone of an operating system over 20 years

old similar to DOS (Unix). That's ALL Linux is. The kernal was taken,
decompiled, and stolen by Linus himself. That's a fact.

That's not a fact - that's utter b***shit. If you even knew a tiny bit
about Linus and the history of Linux, you'd know that.

Linux was written by Linus out of frustrations with Minix. No
commercial Unix kernel was "taken and decompiled" - that's just plain
wrong.

Get your facts straight, Mike, before posting such stupid lies!

Marc
================================================================
Marc Scheuner May The Source Be With You!
Bern, Switzerland m.scheuner(at)inova.ch

 

[_]

Daniel,

I'm going to have to end this interesting discussion, too busy to give it the
attention it deserves. It's been real. Power to the programmers yo!


--
What the caterpillar calls the end,
the rest of the world calls a butterfly.
~ Lao-tzu

Components For Thinkers
www.abderaware.com
zane @at@ abderaware .dot. com

 

[Daniel O'Connell]

That's not a fact - that's utter b***shit. If you even knew a tiny bit
about Linus and the history of Linux, you'd know that.

Linux was written by Linus out of frustrations with Minix. No
commercial Unix kernel was "taken and decompiled" - that's just plain
wrong.

Get your facts straight, Mike, before posting such stupid lies!

Heh, its generally not worth paying much attention to posts like this. The
followup comments can create interesting discussions however.
I should also note, because you didn't and the OP might care, the minix
kernel was free and the subject of a book by its writer. Its quite a good
book actually, however I can't think of the title off hand...have it around
here somewehre.

 

[Daniel O'Connell]

Daniel,

I'm going to have to end this interesting discussion, too busy to give it the
attention it deserves. It's been real. Power to the programmers yo!

Heh, I was wondering who would give up first. It was getting hard to make
responses in a timely manner, I think i spent 30 minutes last night reading
and writing that reply.
It has been interesting.