Minix Case Study?
James
was wanting to know if there was a use for Minix beyond the classroom?
I originally searched google, but yealded few/no real results. The
results was flooded with too much with Linux results.
TIA, James.
sinu.na...@gmail.com
I don't have much idea on case studies on minix, but if you wanted to
know, if we can use minix beyond classes or not, I can say something
about it...
Till minix 2, it was more a teaching tool rather than for general
purposes..If we are planning to use it in yahoo/google servers,
probably it is not the one...but I don't think if there is something
really bad in minix, for which it can't be used as our desktop
purposes! at least for a programmer...It is obviously can't be used as
a Xbox for playing 3D games..the worst case to say...
Minix 3 is rather more to be used as a general purpose OS than a
teaching tool. Many new and exciting things (such as self healing) are
there in it to be as close to commodity OSs as possible and to perform
even better than them, at least in reliability perspective. We have a
GUI (bare X windows system) in it...and a handful of utilities to
serve to our daily needs...For a programmer, it is one of the best
choices.
To know more, we can start playing with it...And more to it, Minix 3
book serves as the manual for it...and we can have a full picture of a
simple and workable OS as a programmer...we can have a bare system to
grow with as a practitioner... and a beautiful tool to learn OS as a
student... and an OS to compare with other OSs to look for more
innovative things to implement as a researcher...
Hope it helps.
Thanks.
Srinu
Barnacle Bill the Sailor
> Concerning the technical aspect of Minix, does anyone know what is
> keeping it out of the server market? What would it take to help meet
> Minix's goal of being used as a general purpose OS? Any ideas would be
> appreciated.
>
Perhaps it is that is only recently promoted for things outside the
educational area? In other words...give it time, it'll be on servers soon.
J.F. de Smit
> Concerning the technical aspect of Minix, does anyone know what is
> keeping it out of the server market? What would it take to help meet
> Minix's goal of being used as a general purpose OS? Any ideas would be
> appreciated.
Device drivers, mainly. Apache, PHP, mysql and the likes have already
been ported, but hardware support is still very limited.
--
Jens de Smit
Student Computer Science | Vrije Universiteit Amsterdam
jfd...@few.vu.nl | http://www.few.vu.nl/~jfdsmit
"[In the end, people] get furious at IT that the goddamn magic isn't working"
-- Stewart Dean
Michael Black
> Concerning the technical aspect of Minix, does anyone know what is
> keeping it out of the server market? What would it take to help meet
> Minix's goal of being used as a general purpose OS? Any ideas would be
> appreciated.
>
it for such things?
Linux is around, and even real Unix. Neither hard drive space nor
ram is a problem these days, so "I need something smaller" is hardly
an issue.
Linux is far more common, which means most of what you want already
exists and it's really easy to find people familiar with it, and find
books and tutorials.
WIth all that, there has to be some incentive to go against that
flow and use Minix. Find that incentive, and you've got your
answer.
Michael
James
I agree Linux is common. More importantly, I think Linux is popular,
because it has support for tons of things and it has a great
development environment. If Minix was to polish its development
environment, do you think developers would come to it? Are there any
developers in this group, that would embrace Minix as a development
platform, if it has support that Linux does?
Michael Black
draw potential users, there is no incentive to change.
How many people stick with Windows? They've paid for it, they're
used to it, it's already installed. There has to be a good reason
for them to give up that comfort to move to another operating system,
which entails a lot of work and learning, and will invalidate
the investment they have in Windows.
There was a time when Minix was the cat's meow. Circa 1990, I
was certainly tempted to spend the $90 or so dollars (more
with exchange) to get the disks for my Atari ST. There wasn't
nearly as clear a field back then as now, there was less of
an investment in any one standard, and it would have given me
"unix" at a far lower cost than the real thing (if the real
thing was even available or within my reach).
But there are alternatives, that are better developed, more
comprehensive. If Minix is comparable, people would still
need a good reason to make the effort to change.
Michael
atman
>
> - Show quoted text -
I agree.Minix doesn't need to be anthoner linux.In educational
erea,its small size is good for us studnents to understand modern
os.In industry,it can find its place in EMBED system.I am a
jackeroo,and I hope minix keeps simple and tiny.
J.F. de Smit
> James (jamez....@gmail.com) writes:
>> Concerning the technical aspect of Minix, does anyone know what is
>> keeping it out of the server market? What would it take to help meet
>> Minix's goal of being used as a general purpose OS? Any ideas would be
>> appreciated.
>>
> Look at it from the other angle, what appeal does it have to use
> it for such things?
> Linux is around, and even real Unix. Neither hard drive space nor
> ram is a problem these days, so "I need something smaller" is hardly
> an issue.
Remember that Minix' small size is only one of the design goals. Also,
this goal is achieved by making the all parts of the OS optional. The
microkernel design and overall structure of the kernel allows for
disabling features unnecessary for a specific purpose. However, once
Minix is "fully grown" I don't think it will be much smaller than
Linux _in its full form_. The power of Minix is that it's easy to
throw anything out that you don't need.
Another important design goal might be more interesting for server
admins: reliability. More and more, servers are the very heart of
almost any company: if a server fails, almost every employee is stuck
until everything is back in working order or a company might miss
revenues due to its website being unavailable. Minix is designed to be
resistant to complete OS crashes, to keep running when other OSes
would require a kick in the head.
Regards,
Jens
Markus E L
James wrote:
> environment, do you think developers would come to it? Are there any
> developers in this group, that would embrace Minix as a development
> platform, if it has support that Linux does?
Or are there any developers at all "in this group"? My impression from
the postings here it, it has become much more quiet than it was, let's
say, five years ago.
Regards -- Markus
Markus E L
On a related tune:
- Anybody knows what's keeping / kept OS/2 from the server market?
- Anybody knows what's keeping Zeta from the desktop market?
- Anybody know what kept BeOS from the desktop market?
- ...
Gaah! People!
After you had thought some time about all this questions, you should
realize that not all reasons are technical and often not even easy to
diagnose. The simplest way to see it (but it's never so simple) is,
that there is already some established solution in market X. Like
Linux or *BSD in example. Or Windows 200x server.
I don't want to bash Minix, but realistically, it will never be ready
for "the server market". Not within the next five years. Apart from
(yet) missing important features for main stream server applications
(therading, virtual memory, power management etc), you also don't seem
to realize how much politics, cost optimization and obscure interfaces
are involved in the server market.
Minix has a really nice code base to start something new. Only what?
Realize that it is very very rare, that a system / application /
whatever replaces another one in a given market in which the other on
is already deeply entrenched. Linux works and has great uptimes --
what's the business case for using Minix in "the server market"?
I don't say all this to malign Minix. But Minix will have to wait
until it's window of opportunity (e.g. a change in the market
landscape) comes. It would be nice to position it against QNX. But
there is already QNX in this market (and it's rather realtime) and
embedded Linux or still NetBSD (to a certain extent). Will be
difficult to grow with so much shadow.
Regards -- Markus
James
or even a BSD, would you embrace Minix? I am posing this question to
the Minix group, because I thought there would be some fans of Minix
in here.
I know you think that it is silly to ask such a question, because of
the obvious. But, remember that Windows and Unix had a share, before
Linux came along. And for some reason, everyone flocks to Linux, when
there are other open source alternatives, to the alternative. While I
think Linux is great technology, I think much of the popularity has to
do with its grass roots movement/evangelism.
Minix3 seems like a great technology, even in its infancy. I really
like the self-healing factor. I just want to know why someone has not
looked to further developing Minix3, instead of the code just
stagnating and/or waiting for some grad students in Europe to crank it
out, in their spare time. (I would love to develop Minix3, but I am
not a OS developer. Maybe one day.)
Rui Maciel
> If Minix was to polish its development
> environment, do you think developers would come to it? Are there any
> developers in this group, that would embrace Minix as a development
> platform, if it has support that Linux does?
I don't know about development environments. I do believe that the lack of
some features (support for USB and SCSI, ELF) makes it hard for some to
maintain their interest. There have been quite a few complains/requests for
those features and in contrast this was the very first time I've seen a
request for a development environment.
Rui Maciel
Markus E L
James wrote:
> No one has answered my question. If Minix has the abilities of Linux,
> or even a BSD, would you embrace Minix?
No. Those other systems already work for me in their designated
area. Changing to Minix would have be pointless. I'd only change if
Minix had a feature I'm sorely missing in the other system.
> I am posing this question to
> the Minix group, because I thought there would be some fans of Minix
> in here.
I'm a Minix fan regardless. But as soon as investement of time and
money is to be considered, fandom alone isn't a factor.
> I know you think that it is silly to ask such a question, because of
No.
> the obvious.
Because of which obvious what?
> But, remember that Windows and Unix had a share, before Linux came
> along.
I'm convinced, strictly speaking, Linux didn't displace Unix, at least
not at the beginning. There was a niche in the market: Inexpensive
Unix on commodity hardware (as opposed to the rather expensive Unix
licenses of the day). Other competitors, like Coherent, didn't make
it, because they lacked networking. And as a hobby OS of the
distinguished CS or engineering student, Linux made the race because
Windows didn't have acceptable networking until Windows 95/98 (the
capabilites of 95 were still rather restricted).
> And for some reason, everyone flocks to Linux, when
> there are other open source alternatives, to the alternative.
^
me: parse error, '.' expected.
$ _
> While I think Linux is great technology, I think much of the
> popularity has to do with its grass roots movement/evangelism.
Only partly. For some time it satisfied an existent need and now of
course it has a user base and a well entrenched base of evangelists.
> Minix3 seems like a great technology, even in its infancy.
Technology needs to be really much much better before being able to
_replace_ existing technology. People have invested (time, money,
knowhow) and are loath to change.
In my opinion the way the change happens, it's that the IT landscape
changes (e.g. form mini computers to personal micros). Then comes a
landslide development where superiority really counts: But only
superiority in satisfying the markets needs. See Microsoft: In a sense
they were successful then, because they took the "personal" in
personal computer serious. Now, of course they are successful because
they are/were successful (meaning: They own the market). DEC on the
other side where successful when the Minis came into the market and
was not flexible enough any more when the micros came, so they went
out of the market.
So the question of course is: What will the next change be?
> I really like the self-healing factor.
Yes, that's cool. It is cool as an academic/scientific exercise. And I
say that with all respect, "academic" for me is a compliment. As
something that might come useful in some niche market. But now tell me
which market that might be, and why Linux (with an uptime of years),
can't serve this market. Note that rebooting or failing over to a
second Linux node is also a valid discovery strategy, so I don't want
to hear 24x7 availability: We can have that already with existing
technology.
> I just want to know why someone has not looked to further developing
> Minix3,
> instead of the code just stagnating and/or waiting for some grad
> students in Europe to crank it out, in their spare time.
This is putting it overly negative.
> (I would love to develop Minix3, but I am not a OS developer.
> Maybe one day.)
Well, you can develop it in your spare time ...
:-) - Markus
Markus E L
Rui Maciel wrote:
> James wrote:
>
>> If Minix was to polish its development
>> environment, do you think developers would come to it? Are there any
>> developers in this group, that would embrace Minix as a development
>> platform, if it has support that Linux does?
>
> I don't know about development environments. I do believe that the lack of
> some features (support for USB and SCSI, ELF) makes it hard for some to
> maintain their interest. There have been quite a few complains/requests for
Yes. Complaints and requests. I wonder why "they" didn't just add the
missing features.
> those features and in contrast this was the very first time I've seen a
> request for a development environment.
Regards -- Markus
Rui Maciel
> Yes. Complaints and requests.
I understand what you mean. It must be a tad frustrating.
> I wonder why "they" didn't just add the
> missing features.
You have to understand that, even between those presently interested in
minix, not everyone is a CS or EE major. Even between those who in fact are
CS or EE majors, not everyone has the know-how and expertise to just get
hold of a keyboard and write a device driver for any operating system, let
alone a fringe one with an "uncommon" design.
If that wasn't the case, naturally that would never be a problem.
But when discussing issues like market acceptance, issues like adoption
barriers are unavoidable. Moreover, no one should expect that those who
gave minix a try and were left underwhelmed due to the lack of some
features should just write their own. Not everyone has the know-how, time,
interest or even motivation to do that job. So why should they be
criticised for not offering any contribution?
Rui Maciel
Markus E L
Rui Maciel wrote:
> Markus E L wrote:
>
>> Yes. Complaints and requests.
>
> I understand what you mean. It must be a tad frustrating.
No, why? I'm not of the Minix team. I also don't know the position of
the Minix team on this, though I'm sometimes wondering (a) why the
even bother and (b) what their long term target is.
I _think_ the long term target is, do produce a more modern version of
Minix, still primarily for _their_ educational and scientific
purposes. But to make the "game" of teaching more interesting, the
story is now that Minix is targetting the embedded sector. I don't
know wether Ben Gras or Andrew Tanenbaum really believe that. I
_think_ they don't. In a sense this might be a pedagogical fiction to
fix the direction of development and discussion (role play for
students: "We have this system and it is targetted to ... Now discuss
...". It's always good to fix some requirements, else discussions with
students drag endlessly ...)
Mind you, this is completely speculative from my side. I think it will
be OK for VU if some interesting thesises and scientific papers spring
from that project and if it get's their students to have a look deeper
into OS theory, into "their own OS".
But the nice thing is, they don't keep the rest of the world from
using Minix for whatever purpose _they_ want. This is great. After
obeserving c.o.m since long before the release of Minix 3, I'd also
say: Don't expect too much leadership from VU (Again: I don't mean
that in a negative way). Just set your own goals and realize them,
using Minix as a basis. As far as I see, the VU team has no deadlines
or near term goals. Quite understandable. So if anybody wants to have
a _product_ (as opposed to "the latest current version") I suggest
they start creating their own distro downstream of VU - In the same
way Linux distros are maintained by pulling together, bundling and
tuning upstream sources.
>> I wonder why "they" didn't just add the
>> missing features.
> You have to understand that, even between those presently interested in
> minix, not everyone is a CS or EE major.
(1) I don't think you need to be a whatever major to be or become a
proficient hacker. The early history of Minix proves that quite a
bit. And Linus T. was only a student, then, BTW.
(2) It's in my opinion more a question of persistence and a do-it-
yourself / help-yourself attitude instead of either believing from
the outset that this is to complicated and or leaving it to others
to do it (or argue that it is their responsibility). To be a bit
ironic: I'm quite sure that, if you're dissatisfied with Minix,
you'll get back every piece of money you paid for it :-).
> Even between those who in fact are CS or EE majors, not everyone has
> the know-how and expertise to just get hold of a keyboard and write
> a device driver for any operating system, let alone a fringe one
> with an "uncommon" design.
Bah! I'm convinced writing drivers for Minix is much easier than
writing them for Linux. Or Windows.
> If that wasn't the case, naturally that would never be a problem.
No, the problem is always not the question how difficult it is to
write a driver, but "What do I want to do, and how do I accomplish it?
Are there faster an simpler ways?". In most cases, the answer to "what
do I want to do" is not "write a device driver", but "I have this
hardware and thought putting a web or shell server on it. But my
networking card is ...". So the question in every single case is: If
Minix doesn't work out of the box as it is for my problem and my
constraints, isn't it cheaper to use some other OS. And the answer in
every single case is usually yes. Of course, except if my problem is
learning about operating systems (and I don't care that the question
in the Kob interview will not be "do you know about operating
systems", but rather "have you ever written a driver for the Linux
kernel"), yes, then I'd use Minix, obviously.
And this is the case of non commericial application. The constraints
are tighter in the commercial sector.
> But when discussing issues like market acceptance, issues like adoption
> barriers are unavoidable. Moreover, no one should expect that those who
> gave minix a try and were left underwhelmed due to the lack of some
> features should just write their own.
No, that's exactly what I wanted to express. One would also not expect
them to claim for others to write those missing features. Why should
VU? And they _have_ an interest in Minix.
The only thing you can suggest at this point is how happy a huge
number of people were if popular feature / network card etc. would be
supported. And hope the missing feature bugs the VU enough so that
they become susceptible to the suggestion.
Everything else is unrealistic. I can't believe VU aiming for "world
domination" ;-).
> Not everyone has the know-how, time, interest or even motivation to
> do that job.
> So why should they be criticised for not offering any
> contribution?
No, I criticize them for complaining as if they are paying customers,
somehow insinuating that VU should be interested that they (those
people) run their (VU's) Minix. But that is just unrealistic.
On the other side I'm quite sure that one can try to pay a grant to
finance a thesis which undertakes a certain extension of Minix. But
I'm quite sure that was never the question in those cases.
In general: I don't care wether people complain on missing features or
ask naive questions on why Minix hasn't (conquered the X market yet or
missing a feature or whatever). They are entitled to ask.
The only _observation_ I want to add to those requests is, that to me
they seem to indicate rather a lack of understanding how "the market"
works. And from that, probably, all short answers to those questions
will be unsatisfactory ("Why don't you just add, do this or that ...,
I'm sure then ...?").
But as I already said: I've nothing to do with VU, at the moment only
rarely browse over the Minix3 site and c.o.m has certainly become much
more silent (actually I'm missing Kees Bots explanations), and,
frankly, the development process of VU and it's aims are rather opaque
to me.
If anybody wants more to happen, I suggest a community Minix project
:-). But I haven't got the time for the next year(s), so don't count
on me.
Regards -- Markus
Michael Black
>>> I wonder why "they" didn't just add the
>>> missing features.
>
>> You have to understand that, even between those presently interested in
>> minix, not everyone is a CS or EE major.
>
> (1) I don't think you need to be a whatever major to be or become a
> proficient hacker. The early history of Minix proves that quite a
> bit. And Linus T. was only a student, then, BTW.
>
> (2) It's in my opinion more a question of persistence and a do-it-
> yourself / help-yourself attitude instead of either believing from
> the outset that this is to complicated and or leaving it to others
> to do it (or argue that it is their responsibility). To be a bit
> ironic: I'm quite sure that, if you're dissatisfied with Minix,
> you'll get back every piece of money you paid for it :-).
>
And there is a certain irony here. An OS is created to serve as
an example when teaching about operating systems, people jump on
the bandwagon to use it beyond that purpose, and then expect features
from the teachers.
So those who want to "corrupt" the OS into something more than
a teaching tool have "corrupted" it further by not using the
resource to learn and extend the OS.
It seems that Linux came about because there was Minix, though
it's not clear to what extent it actually served as a teaching
tool and what extent it was "merely" a launching platform for
the early work on Linux.
That's a pretty good lesson.
Michael
Barnacle Bill the Sailor
Unix like system to use on x86 of which he had one. He was at the time
studying Minix but it was not ported to x86 or lacked some features that
he wanted. So yes, Linux arose from Minix. When Linus posted here for
advice the ball got rolling, and I think I can say the Linux momentum
has pushed Minix along as well. Certainly the open source movement in
general has benefited directly from it all. Although I wonder if GNU
hurd has not seen much sunlight because of it.
Markus E L
Barnacle Bill the Sailor wrote:
To put this a bit more precise: Minx has been written for the x86
originally, though it was the 8086. Later versions hat protected mode
support (80286), but there never was a main line 32bit Minix up to (I
think) the recent 3.x series. Ther were patches (Bruce Evans Minix 386)
and Minix-VMD (derived from 1.7) has also 32bit. The problems were twofold
- Minix could only be distributed as patches at the time, since
Prentice hall held the right to the source (IMHO it must have
seemed as a sensible arrangement at the time when there was not
network access for anyone so the "free donwloading" didn't come
into the equation anyway).
- If understood Linus right (in his book "Just for Fun"), he began to
write his kernel because he was dissatisfied with the terminal
emulation in Minix. Later he needed to download with the terminal
emulater, so he added a filesystem etc. Initially he used Minix as
a development platform (so it was workable ... :-), at least).
(Actually, now that I write it it seems inbelievable: Instead of
writing a proper terminal emulator for Minix, the man writes a
terminal software running on the bar metal, then extends it to an
operating system. Of course he also wanted to know about programming
the 386 CPU really low level, that explains it, but the irony is still
there, since he starts with low level programming and over the years,
again we get a _portable_ operating system (which means, Andrew
Tannebaum was essentally right in the famous flame war)).
> So yes, Linux arose from Minix. When Linus posted here for
> advice the ball got rolling, and I think I can say the Linux momentum
> has pushed Minix along as well. Certainly the open source movement in
> general has benefited directly from it all.
> Although I wonder if GNU hurd has not seen much sunlight because of
> it.
Oh well. This is another long and convoluted story ...
Regards -- Markus
Markus E L
'et472 AT FreeNet DOT Carleton DOT CA (Michael Black)' wrote:
>> (2) It's in my opinion more a question of persistence and a do-it-
>> yourself / help-yourself attitude instead of either believing from
>> the outset that this is to complicated and or leaving it to others
>> to do it (or argue that it is their responsibility). To be a bit
>> ironic: I'm quite sure that, if you're dissatisfied with Minix,
>> you'll get back every piece of money you paid for it :-).
>>
> And there is a certain irony here. An OS is created to serve as
> an example when teaching about operating systems, people jump on
> the bandwagon to use it beyond that purpose, and then expect features
> from the teachers.
Oh yes. Irony abounds :-).
Regards -- Markus
muchan
> Bah! I'm convinced writing drivers for Minix is much easier than
> writing them for Linux. Or Windows.
When a driver for certain device is written for Linux as open source project,
how dificult is it to port it to Minix?
(Well, I didn't yet to see a line of source of either Minix nor a driver for
Linux...)
muchan
Michael Black
drivers in one of the manuals that came with the OS. When people needed
drivers for different hardware, they'd use one of the samples as the basis of
the rest. It helped a lot.
On the other hand, the problem with writing drivers is often not the
coding, but in getting the information about the hardware. The fancier
the hardware is, the more complicated the driver needs to be. And many
boards are not issued with the needed information, the companies want
you to use their driver (for Windows) and that's it. So in order to
figure out how to control the board, one has to do a lot of work to
get that information.
Michael
Markus E L
muchan wrote:
> Markus E L wrote:
>> Bah! I'm convinced writing drivers for Minix is much easier than
>> writing them for Linux. Or Windows.
>
> When a driver for certain device is written for Linux as open source project,
> how dificult is it to port it to Minix?
I don't think the activity could be called porting :-). Actually it's
more: Write another driver after you've had some experience with the
hardware. That is to say: AFAICS the environment in which a Minix
driver runs is very different from the environment a Linux driver
uses. Straight porting is out of question, IMHO, though one could read
a Linux driver and learn something about corner cases etc. which would
be useful to write a driver for Minix - if the Linux driver is well
commented.
Others might correct me: Please do so.
Regards -- Markus
muchan
> muchan (muc...@promikra.si) writes:
>> When a driver for certain device is written for Linux as open source project,
>> how dificult is it to port it to Minix?
>>
>> (Well, I didn't yet to see a line of source of either Minix nor a driver for
>> Linux...)
>>
> When I ran Microware's OS-9 twenty years ago, there were some sample
> drivers in one of the manuals that came with the OS. When people needed
> drivers for different hardware, they'd use one of the samples as the basis of
> the rest. It helped a lot.
>
> On the other hand, the problem with writing drivers is often not the
> coding, but in getting the information about the hardware. The fancier
> the hardware is, the more complicated the driver needs to be. And many
> boards are not issued with the needed information, the companies want
> you to use their driver (for Windows) and that's it. So in order to
> figure out how to control the board, one has to do a lot of work to
> get that information.
>
> Michael
>
Thank you for answering.
I thought all the "information" needed are in the open sources of Linux drivers.
muchan
muchan
Thank you for answering.
I thought if the Linux drivers source is well structured to layers of
- basic communication with hardware
- protocol layer
- os dependent code
- maybe machine dependent part of same os
the first two might be ported/used without a lot of modifications.
If a few students are asigned to write a missing driver for common hardwares,
may be in a few years Minix would support much more hardwares? :)
muchan
J.F. de Smit
Technically, you are correct. However, some programmers do not like to
comment their code ("If it was hard to write, it should be hard to read"),
resulting in very difficult reverse engineering for the person who wants to
rewrite the driver for another system...
Michael Black
>
> Technically, you are correct. However, some programmers do not like to
> comment their code ("If it was hard to write, it should be hard to read"),
> resulting in very difficult reverse engineering for the person who wants to
> rewrite the driver for another system...
>
I was looking at some code for a Linux application last
night, and it was nicely broken down into sections. But
while there were intriguing names to the sections, there
was no preamble about what the section was supposed to do.
It just seemed like "carelessness".
Now, I barely have a grasp of C, so what might be obvious
to others is not obvious to me. But I used to disassemble
programs years ago in order to fix something or just to see
what was going on, and it took a lot of work to figure out
what was going on, because while it was obvious that
instruction there was doing such and such, some of the
information was missing, the bigger picture. The same with
commented code, if I'm not sure what this section is doing,
then I have to work hard to figure out what it's doing. It's
kind of like Morse Code, which is really easy to send, but that
doesn't actually help you in being able to receive it
COmmenting code is really an art. I suspect many do it
mainly because they "know they should", but since they
don't really know what they should comment (and at the time
they are adding the comments, they know the code well enough
that they think they'll always remember what they are doing),
then the comments are haphazard. What they think is important
to comment may not be what others need in the way of comments
to follow the code.
Someone commenting code to teach is bound to put a lot more care
into what they comment, and the comments are as important, if not
more important, than the actual code. They are trying to think of
what the reader is seeing, a key to any teaching.
If they don't embed the information about the hardware at the
beginning, then the lines of comments in tandem with the code won't
help much. Because the information about the hardware is the
preamble to the actual code.
Michael
Markus E L
muchan wrote:
Even then I'd doubt it's so easy, and generally I don't believe that
Linux drivers are this well structured. But even if they were, I'm
convinced that you can't just take over their pieces in a simple way:
Minix drivers are single treaded and if they want to serve multiple
longer transactions they must multiplex (by which I mean, store the
state associated to a transaction somewhere and poll for further
messages during a transaction). Linux on the other side is
multithreaded in the kernel too (but cooperatively). Maybe those two
different models (and the primitives used to control the drivers in
Linux vs. those in Minix) can be easily mapped to each other. Or maybe
they don't. Maybe this would be a useful research project.
(Again: Anybody who knows better, please correct me: All this is what
I gathered only from reading Minix 2 source long years ago and from
digging a bit in the "but the Minix FS is a bottleneck because it's
singlethreaded" debate -- still a non sequitur to me).
Generally reading code is useful, though, and one might learn from
reading _some_ Linux drivers. But this all depends on the efficiency
with which one is able to read foreign code, which varies wildly even
between fluent programmers. But be prepared to spend vastly more time
than expected. And as others already said: The major stumbling stone
will be the lack of technical documentation WRT hardware, and to
extract this from the Linux drivers - good luck :-)
> If a few students are asigned to write a missing driver for common hardwares,
> may be in a few years Minix would support much more hardwares? :)
Ask VU :-). But I don't think they see their students as a ressource
to be exploited to push Minix into some market. Rather the other way
round: Science and education comes first (and opening Minix since 1.7
upwards has, I suspect, probably been done to further those ends, not
to get anybody out there a cheap Unix OS, though this, of course might
be a welcome side effect).
Regards -- Markus
Rui Maciel
> And there is a certain irony here. An OS is created to serve as
> an example when teaching about operating systems, people jump on
> the bandwagon to use it beyond that purpose, and then expect features
> from the teachers.
That may have been the case of minix 1 and 2 but minix3's goals cover a bit
beyond that. From the site:
"MINIX 1 and 2 were intended as teaching tools; MINIX 3 adds the new goal of
being usable as a serious system on resource-limited and embedded computers
and for applications requiring high reliability"
So naturally, as the target audience increases, you get "people jumping on
the bandwagon" who lack the technical skills to even write a "hello world"
program. You cannot expect drivers and servers from those potential (and
not so potential) users. Someone else must pitch in because if no one else
does, they sure can't do it themselves.
Rui Maciel
fatpig
primary goal while tentatively extending a tentacle into a wider
domain such as embedded systems, servers etc., I find it something of
an oversight that MINIX 3 doesn't have at least enough USB support to
use common peripherals as standard. Yes, that would make things more
complex from an educational perspective but that's just the nature of
the beast these days. MINIX 3 with no USB support will soon be what
the original MINIX would have been like in 1987 with no floppy or
serial driver. Why do I talk about USB? Beacuse many (most?) new
machines have neither a serial port or floppy drive. Soon you won't be
even able to buy floppy discs or they'll become an expensive
speciality. How does a person with a fairly recent machine use his
USB keyboard? How does he transfer files with his 2/4/8GB memory
stick? He can't with MINIX. This is a major impediment even if the
objective is "only" for education. It's one thing not to burden the
system with lot's of unnecessary "bells and whistles", it's quite
another not to have basic functionality with modern hardware.
As for the people who say "if you want USB support write it yourself",
well I would definitely like to. I consider myself an upper-
intermediate level C programmer with the ability but in common with
most, finding the time would be a major problem. That's no idle
dismissal or a cop out. I spend most of every working day at work and
have a wife and family to give attention to.
For the record, I've been intermittantly playing around with MINIX
since about 2000 and I'm not one for upgrading for the sake of it. I
have several '386, '486 and early Pentiums which run just fine. But
I'm not the common user. Most people want at least a fairly recent
machine and these machines are taking the form mentioned above.
Best Regards,
Charles L
Philip Homburg
fatpig <cwl...@tpg.com.au> wrote:
>the beast these days. MINIX 3 with no USB support will soon be what
>the original MINIX would have been like in 1987 with no floppy or
>serial driver. Why do I talk about USB? Beacuse many (most?) new
>machines have neither a serial port or floppy drive. Soon you won't be
>even able to buy floppy discs or they'll become an expensive
>speciality.
A small team cannot do everything at the same time. Eventually, a USB
implementation will arrive...
>speciality. How does a person with a fairly recent machine use his
>USB keyboard?
Fortunately, this is not a problem. The BIOS (in SMM) will deal with this.
--
That was it. Done. The faulty Monk was turned out into the desert where it
could believe what it liked, including the idea that it had been hard done
by. It was allowed to keep its horse, since horses were so cheap to make.
-- Douglas Adams in Dirk Gently's Holistic Detective Agency
d...@kbrx.com
port converters are available from numerous suppliers. mpja.com (u.s.
company) has them at US$1.95 each. USB to rs232 converters are also
available somewhere. All handy for old computers.
Hul
Rui Maciel
> PS keyboard to USB computer port and USB keyboard to PS computer
> port converters are available from numerous suppliers. mpja.com (u.s.
> company) has them at US$1.95 each. USB to rs232 converters are also
> available somewhere. All handy for old computers.
Those connectors aren't available everywhere. In fact, the only converter I
ever saw in any store in my local area was for USB mice. I really looked
for those things, as I had purchased an old computer which only supported
USB peripherals and I only had PS/2 keyboards.
I believe that hardware stores don't try very hard to offer those connectors
due to the fact that if they aren't available then the only option that the
client has is to buy a new keyboard. Between selling a 2€ and a 10€
product, they prefer the later option.
Rui Maciel