So my question to the Slashdot community is when is Linux going to be prevalent enough on the desktop that people will pay for applications and not always assume they are free?
You seem to miss the whole point of free software.
My question would be:
when is Linux going to be prevalent enough
on the desktop that people will
always assume that software should be free?
Think about it -- the answer will surely
be somewhere between your question and mine.
If you want to sell proprietary software
working under free software operating systems,
which exist only because some people have
rebelled against proprietary software world,
then you will have to find out which
of our two questions the answer is closer to.
Better yet, where are the people who feel that way now?"
They are probably running Windows or Mac.
I don't really think people who use
totally free software
operating systems
-- be it GNU/Linux, GNU/Hurd, FreeBSD,
OpenBSD, NetBSD, etc. -- on their desktops,
which, mind you, still often means
many disadvantages
in cooperation with
proprietary software world
(e.g.
Microsoft Office obfuscated documents format),
would like to use your proprietary software
(which has nothing to do with paying for it),
because such a software is
exactly what they are trying
to be free from.
I am using only free software today
(and I have paid lots of money for some of it)
on my servers, as well as on my desktops,
and this is how does it look like for me:
running Debian GNU/Linux with Apache and mod_perl
plus PostgreSQL or MySQL is easy --
as long as it produces HTML, we are compatible
with the rest of the world
(even Microsoft browsers under Microsoft OS
will have no problems at all)
-- but using GNU on the desktop can be much
more problematic.
There are websites which use ActiveX,
there are people who send MS Excel or
MS Word documents
in email
(even if it's HTML written by MS Word,
when I look at the source code,
I don't know if I should laugh or cry),
there are people who send
different presentations and other data
in the form of Microsoft Windows executables,
et cetera, et cetera, every day there is some
problem.
So, why do I use free software on
my desktop, you may ask me?
Because I strongly believe in my freedom,
which I value much higher than
my convenience.
This is why I will never use your proprietary
software, and this is maybe why some
of other people will never use it as well.
Of course, I can speak only for myself,
but I believe other people, who use exclusively
free software desktops, do so because of
different reasons than
their convenience, as well.
And this is why I would suggest you
targeting proprietary software to people who
use proprietary software, which, after all,
sounds quite obvious,
but this might be exactly your problem here.
If you want to target software to me,
or to people like me,
take a look at
GNU Philosophy
and
Debian Social Contract
to have some idea how do I choose my software.
I wish you good luck.
I surely don't, otherwise I wouldn't
have written them, now would I?
But I hope you are about to enlighten me.
So it's cleaner to make the maximum filesystem size depend on the machine's pointer length than it is to simply write better software, even if writing better software means resorting to traditional ("legacy" as you might call it) methods?
Yes, it is a cleaner interface.
It might be not enough in the terms of
partition size on 32-bit architectures
for some tasks, but it is still
a cleaner interface.
The "single opcode MOV?" You do realize how mmap works?
As a matter of fact, yes I do.
If I didn't, I wouldn't have
been writing about it,
if it isn't obvious.
I surely hope that I was wrong and
you will now proceed to teach me the
mysteries of mmap, since
I'm getting a little bit tired of
your aggressive rethoric.
Certainly the user-level program can simply access it like memory,
And which part of my post has proved you
that I thought otherwise?
(Of course you do know that
the Hurd servers are user-space processes,
don't you?)
but this then goes through the VM (by means of a hardware interrupt on page fault) which calls some disk subsystem, which will eventually end up calling a disk driver. End result is that you write something to the disk.
Dear Lord!
Please! Have you really though that I
had no idea that the whole point of
having an interface to partition is to
write something to the disk?
Have you really thought that
in my eyes it was some magical way to make
the hard drives not needed any more?
This is an insult, no more, no less.
There is no magic MMU hardware which understands EIDE or SCSI and maps memory to disk blocks - this is done through code. It's merely an abstraction to transfer the complex parts of the code into another layer of software, made necessary because of the poorly-thought-out layers of software.
Of course it's merely an abstraction,
this is what interfaces are all about.
I think you have totally misunderstood my point
-- I was talking about the
interface, not about the
implementation.
How can a single MOV be considered "elegant" when you have to keep a data structure of which parts of the partition are mapped? Your single MOV instruction becomes a series of instructions to select a "view" of a partition, clean out old "views," and map new ones when needed. It won't be a MOV, it will be a function call into some library.
Yes, you definitely seem to talk about the
implementation, while trying to fight my
arguments about the interface.
There's nothing I can say here, since
it has nothing to do with the
fact if the interface is clean or not.
It's not "clean" or "elegant" in any sense of the word.
There is a difference between
"elegant implementation"
and "elegant interface."
The mmap is a nasty hack because someone couldn't figure out how to use the same cache management that the VM uses in a user-level process. Someone simply chose a poor abstraction. It's a mistake, and I simply cannot understand how someone could let such a stunning error into any operating system.
You seem to get it very personally,
it almost seems like you are the inventor
of the open/seek/read/write/close interface.
The use of such an interface in the Hurd,
i.e. making the partition
appear as a memory block,
has nothing to do with the cache.
It finally allows to use it like just another
kind of memory storage, as
the hard drives are all about being a huge
array of bytes.
The hard drives hardware interface start to
look more like just an array of bytes
(in the oposition to heads,
cylinders and sectors),
which I personally find very useful,
and I'm glad the software interface is
going in the similar direction --
i.e. much cleaner interfaces hiding the
actual implementation, which is often
much more complicated.
This is the object oriented way of designing
systems.
I hate to spoil your naivete,
You surely are rude, I must say.
but Intel and AMD don't give a damn about HURD, and they certainly don't design their hardware with this sort of thing in mind.
First of all,
what made you so sure that I thought
Intel (or AMD for that matter)
"gives a damn" about the Hurd?
Second of all,
do you suggest that if Intel or AMD
don't consider the Hurd an important
project
(if this is what "not giving a damn" means,
please forgive me if I don't follow the street
slang)
then it must be because of the fact that
its internal partition access interface
is not clean enough?
(The Hurd is still POSIX
(as well as ISO, ANSI, BSD, Single Unix, SVID, X/Open) compatible, mind you.)
I'm sorry but I'm afraid I can't follow
your amazing logic.
Although the theoreticians don't like this, computer hardware is complex and non-orthoganal. If the HURD was designed to run on only 64-bit machines with clean MMUs, you'll be waiting a long time until HURD is stable on readily-available hardware.
Now you try to say that the
Hurd will work best on better architectures
than the most popular mass-produced hardware
we have today,
which proves
"how ridiculous [my] words
[about the partition access interface
being clean]
are,"
as you kindly pointed out,
or am I missing something?
(Does HURD even support 64-bit architectures at this time? If not, then how can you possibly justify the mistake of having filesystem drivers use mmap?)
The Hurd, as I'm sure you already know,
supports those architectures which are supported
by the underlying microkernel the Hurd itself
runs on top of.
I don't give a damn about the politics.
I don't know what scares me more --
a person who don't care about politics,
or someone who say so but
conclude his arguments with words
"But anyway, mmaping an entire partition is really nuts. They're not getting any sympathy from me."
-- fortunately,
your (or my, for that matter)
political ignorance
shouldn't be important
in our argument
(or at very least I hope so).
This is simply bad design. You can attack me with all sorts of accusations that I don't approve because of whatever politics,
I really don't know why are you so afraid of being accused of politics influencing your
arguments.
Have
my answer to your post
really made you think that I am
suggesting that?
Please point out the appropriate part,
I'd like to know which part
of it has offended you so much
(for which I'd like to apologize you
in advance, since that was no my intention
whatsoever).
but this does not change the fact that someone made an immense error when designing the filesystem servers and certain people thought this error was "elegant."
It seems like those "certain people,"
some of which have been working on
the OS kernels architectures for decades,
could learn a lot from you.
I strongly suggest you to contact them
and make use of your knowledge.
You can find some of them working at
Carnegie-Mellon University,
Berkeley, or
Massachusetts Institute of Technology
in Cambridge.
The Artificial Intelligence Laboratory
on MIT would be a good start. Good luck.
I really didn't mean to flame you personally,
Well, if you didn't mean to "flame" me,
than why on Earth have you done it?
(I ask, because such a rude attitude
doesn't make
our discussion about partition access
interfaces any easier -- and any nicer as well.)
but this is inexcusably bad design. I don't understand how someone can justify it, rather than admitting that it's a fundamental design problem.
Have you ever stopped to think that maybe you are the one who is wrong here?
You can't have cleaner interface to a partition than seeing it as a continuous memory block
Maybe, but it takes exactly fifteen minutes to do:
read_partition(unsigned long start, unsigned long length, char * buffer);
write_partition(unsigned long start, unsigned long length, char * buffer);
Which gives you just as clean interface, if at the cost of a few more calls.
Few more calls and the time to read the whole
partition into the memory, change few bytes
while blocking the partition to any other access,
and then write the whole partition to the disk.
The whole point of using mmap-kind of
disk-memory relationship is not to cache
the partition in RAM,
but to make the
partition appear like it was in RAM
(see man mmap).
Yes, I did. I installed it when Debian first made its distribution public last year.
I threw it out shortly after the first crash.
Such a large project as an operating system,
which
"is not officially released yet, and won't be for some time"
(a quote from
Debian GNU/Hurd website, I linked to in
my post)
is not
totally useless if it ever crashes,
especially if it's soon after the first public
release. Have you tried Linux 0.2 lately?
Let me quote a little bit more from the website:
The Hurd is under active development, but does not provide the performance and stability you would expect from a production system. Also, only about every second Debian package has been ported to the GNU/Hurd. There is a lot of work to do before we can make a release.
Until then, you can participate in the development if you want. Depending on your experience and time commitment, you can help us in many different ways. For example, we need experienced C hackers to develop and implement new features and to fix bugs and debug the system. If you are not very experienced in C programming, you can still help: Either by testing the existing systems and reporting bugs, or by trying to compile some unported software you have experience with. Also writing documentation is important, or maintaining the web pages.
Well, it doesn't look as a production-ready,
bug-free system to me, and it's a
year after you have tried it,
as you said.
Seriously, did you expect
a rock solid mature operating system
from such a description?
Only a clueless academic regurgitator such as yourself will think that when HURD gets a couple MORE of years of development under its belt, that it will run more quickly than the linux kernel.
Please pardon me if I take offence to
your insults.
I'm a little disgusted when people start to use
terms like
"clueless academic regurgitator such as yourself"
to desperately defend
their invalid arguments.
I have NOT said that Hurd
would
ever run more quickly than Linux,
I have really no idea why have you made it up.
Let me get this straight:
The Hurd will NOT run faster than Linux.
(I'm not affraid to agree with you, even after
you called me a "clueless academic regurgitator" --
I try to not take your insults personally,
for the sake of our argument.
Please try to understand, if it is sometimes
hard to me, I quickly get disgusted when
people, sometimes much younger people,
use insulting language.)
It's actually very simple:
the more you throw into the kernel space,
the faster OS you have.
This is why Linux is faster than Hurd,
and this is also why Windows has faster graphics
than Linux.
But also, the more you have in kernel space,
the larger portion of your OS has to be
made secure, because every bug can
potentially crash or exploit the system.
Or HURD will run more stably with all those system and application threads competing with each other to get work done. Oh, people will be running on top of each other to run a slower, buggier OS.
And here you are wrong.
Those new layers of abstraction,
the user-space Hurd servers processes,
making the kernel-space as small as possible,
et cetera -- it's all being done
to make it more stable and
secure, with cleaner design, even
if it is slower than the more monolithic
kernel design
(and here by monolithic I mean the whole kernel
code running as a privileged kernel-space code
with no protection between modules
--
see kernel-level rootkits for example
why this may be dangerous
--
not just the kernel with one binary file).
It may be slower, but
not buggier.
The difference between Hurd and Linux
(and Linux and Windows, for that matter),
is quite similar to the difference
between Lisp and C (or C and Assembly),
i.e. it can be slower, but this is the price you
pay for potentially much more stable design in
the long run.
This is why you should consider the differences
in Hurd and Linux design as a different
direction of speed-stability trade-off.
Of course Linux 2.2 or 2.4
(or even 2.5)
is more stable now than Hurd 0.2
which shouldn't be surprising at all.
And people love to spend their free time developing for obscure platforms like xBSD, QNX, and BeOS
If you don't want to develop for any platform
you don't like, you just don't have to,
I thought it should be obvious.
But what I would suggest you is learning how to
write portable code.
Remember that Hurd is being developed
with respect to such standards as
ISO, ANSI, POSIX, BSD, Single Unix, SVID,
X/Open (and maybe others which I don't
remember about now).
GNU people did a great job at providing
tools which help writing portable code
(GCC, GNU Make, GNU Automake,
GNU Configure, etc.),
but there are also many other tools as well.
Look, maybe in your world RMS is God, and believe everyone should run their lives by his dogma.
(...)
But you and your smelly, unwashed zealots missed the whole point of my exposition.
I'm sorry, but I affraid I will not be able to
continue this discussion, unless you
try controlling yourself.
I know this is Slashdot, but please let us
at least try to not be so childish.
No commercial entity is going to run their webservers or databases on HURD in this decade (if ever). They will chose a commmercial product or chose the OS with the faster, more stable kernel (it will not be HURD).
Of course it will not be Hurd 0.2,
like it was not Linux 0.2.
But please remeber that
"the faster" and "the more stable"
don't have to be the same kernel.
I would bet that in the long run
Linux will be faster, and Hurd will be
more stable.
The same goes for people who use their computers to websurf, write documents, and handle email. These are applications used by the "real world".
You are right that people
(by which I mean most of people)
don't use Hurd, Linux (and GNU at all),
BSD, etc.
-- they use Windows for that.
But should the conclusion be that therefore
Windows design is superior to Hurd?
I don't think so.
After all, more people eat at Mac Donald's
than any other restaurant, which doesn't
mean it's because they serve the
most extraordinary dishes.
The value of HURD is not as a viable alternative to Linux. HURD as valdidation of RMS communist theories is only of value to RMS.
I'm sorry, but this time you have gone too far.
If your motivation for this argument is
insulting Mr. Richard Stallman
(or any one else for that matter,
even people who I personally don't agree with),
then I can no longer stay patient.
I can handle personal insults in my
direction, but insulting other people is
definitely not in my nature.
Its not an alternative to Linux. Its an orange to Linux's apple. It will suck as an alternative to Linux. (...)
Very few people will want to port useful packages to HURD;
(...)
HURD's purpose is not a platform to run applications.
Have you ever heard about
Debian GNU/Hurd?
Are the four (yes, 4) full CDs*today* just
"few people [porting] useful packages" to you?!
Truely amazing troll --
and modded as Score:5, Insightful --
my congratulations, you have successfully fooled
the Slashdot moderators.
Now please don't spread this lies,
because this crap is nothing more than
just lies. Maybe next time before you post
something like this download Debian GNU/Hurd
ISOs, burn them, install the system, count
how many packages there are, and then
post your opinion about it.
This is completely nuts.
(...)
using mmaping an entire partition is just crazy. This is poor design. What were they thinking?
(...)
Make two new system calls, say "readaddr" and "writeaddr".
More syscalls is not the answer
to "poor design," as you call it,
I would call your
open/seek/write/read/readaddr/writeaddr/close
software interrupts based
syscalls interface a poor design,
in the oposition to MOV
fundamental processor opcode,
which is as simple and clean as only
it is possible.
Besides, Hurd is not about speed --
they don't do it to eliminate
a seek syscall overhead
(which is laughable, by the way,
compared to read and write syscalls)
-- it's about clean and robust design.
I, for one,
am sure they are doing a great work,
I personally haven't seen a better kernel design yet.
Another possible approach: using mmap is nice, but mapping in an entire partition is fubar. Why not map in specific parts of the partition as they become needed?
This is what they would probably end with, as
a workaround for poor legacy
32-bit architectures, like Intel x86 --
but only as a quick-and-dirty,
temporary workaround.
You can't have cleaner interface to a partition
than seeing it as a continuous memory block.
At least I can't imagine a cleaner interface,
if you have any idea of such a thing, I'll be more
than glad to hear it.
But anyway, mmaping an entire partition is really nuts. They're not getting any sympathy from me.
It's not a popularity contest
-- it's a clean design, no more, no less.
So in other words, when the GNU project finally produces a complete operating system with all the operating environment trimmings, the correct name will be "The GNOME/GNU/Hurd/Mach System"?
Will they change the name?
Yes, they will change the name to GNU --
finally ending this whole farse of
such a lame jokes like yours,
as well as this
childish ego-masturbation of Linus.
Is this the answer you were expecting?
I hope it will be able to run the new Mosaic software. Have you guys seen that? It's like Gopher but with you can add pictures, change the font size, etc.
Great joke!
It's so funny, because GNU Hurd design is so
outdated! No innovation at all!
Now, Linux -- that's what I call
innovation!
Who the hell modded it up?
is it just me, or does it sound like they had it all ready to ship, date planned and everything, and then someone pointed out that it was lacking some major I/O features/performance, and the developers collectively slapped their foreheads and went "oh shit, yeah, we kinda forgot about that one."
Yeah, you're right,
what a stupid bunch of loosers! Ha ha ha!
like, all this took them by surprise? sucks to forget to implement a couple crucial features, eh?
Yes, sir! Well said, sir!
I suppose you are working on a better
revolutionary
mikrokernel-based system of servers?
Now, where's its URL again?
Oh, I'm sorry,
you mean you are just bitching
because you personally don't like Richard
Stallman and you really have no idea
what the GNU Hurd is really all about
in topics which involves more intellect than
just counting how many gigabytes of filesystem
you can mount?
Please tell us then how would you suggest
mapping multi-GB partitions on a 32-bit
architecture?
Oh, you had no idea that it's being mapped
as memory instead of the old
read/write/seek interface?
To moderators:
Yeah! It's so (Score:5, Interesting)!
You're great!
Should you mod me down as flaimbait?
Of course, since I'm answering to
an obvious troll.
Slashdot Mirror
You seem to miss the whole point of free software.
I am one of people who will pay for applications and always assume they should be free, by which I mean that they should give me enough freedom -- see The Free Software Definition, What do you mean by Free Software?, or The Debian Free Software Guidelines (or The Open Source Definition, which is basically Debian Free Software Guidelines rewritten).
My question would be: when is Linux going to be prevalent enough on the desktop that people will always assume that software should be free?
Think about it -- the answer will surely be somewhere between your question and mine. If you want to sell proprietary software working under free software operating systems, which exist only because some people have rebelled against proprietary software world, then you will have to find out which of our two questions the answer is closer to.
They are probably running Windows or Mac.
I don't really think people who use totally free software operating systems -- be it GNU/Linux, GNU/Hurd, FreeBSD, OpenBSD, NetBSD, etc. -- on their desktops, which, mind you, still often means many disadvantages in cooperation with proprietary software world (e.g. Microsoft Office obfuscated documents format), would like to use your proprietary software (which has nothing to do with paying for it), because such a software is exactly what they are trying to be free from.
I am using only free software today (and I have paid lots of money for some of it) on my servers, as well as on my desktops, and this is how does it look like for me: running Debian GNU/Linux with Apache and mod_perl plus PostgreSQL or MySQL is easy -- as long as it produces HTML, we are compatible with the rest of the world (even Microsoft browsers under Microsoft OS will have no problems at all) -- but using GNU on the desktop can be much more problematic. There are websites which use ActiveX, there are people who send MS Excel or MS Word documents in email (even if it's HTML written by MS Word, when I look at the source code, I don't know if I should laugh or cry), there are people who send different presentations and other data in the form of Microsoft Windows executables, et cetera, et cetera, every day there is some problem.
So, why do I use free software on my desktop, you may ask me? Because I strongly believe in my freedom, which I value much higher than my convenience.
This is why I will never use your proprietary software, and this is maybe why some of other people will never use it as well. Of course, I can speak only for myself, but I believe other people, who use exclusively free software desktops, do so because of different reasons than their convenience, as well.
And this is why I would suggest you targeting proprietary software to people who use proprietary software, which, after all, sounds quite obvious, but this might be exactly your problem here. If you want to target software to me, or to people like me, take a look at GNU Philosophy and Debian Social Contract to have some idea how do I choose my software. I wish you good luck.
To moderators: Yeah! It's so (Score:5, Interesting)! You're great! Should you mod me down as flaimbait? Of course, since I'm answering to an obvious troll.