From: torvalds@klaava.Helsinki.FI (Linus Benedict Torvalds)
Subject: Apologies (was Re: LINUX is obsolete)
Date: 30 Jan 92 15:38:16 GMT
Organization: University of Helsinki
In article <1992Jan29.231426.20469@klaava.Helsinki.FI> I wrote:
>Well, with a subject like this, I'm afraid I'll have to reply.
And reply I did, with complete abandon, and no thought for good taste
and netiquette. Apologies to ast, and thanks to John Nall for a friendy
"that's not how it's done"-letter. I over-reacted, and am now composing
a (much less acerbic) personal letter to ast. Hope nobody was turned
away from linux due to it being (a) possibly obsolete (I still think
that's not the case, although some of the criticisms are valid) and (b)
written by a hothead :-)
Linus "my first, and hopefully last flamefest" Torvalds
From: pmacdona@sanjuan (Peter MacDonald)
Subject: re: Linux is obsolete
Date: 1 Feb 92 02:10:06 GMT
Organization: University of Victoria, Victoria, BC, CANADA
Since I think I posted one of the earliest messages in all this discussion
of Minix vs Linux, I feel compelled to comment on my reasons for
switching from Minix to Linux. In order of importance they are:
1) Linux is free
2) Linux is evolving at a satisfactory clip (because new features
are accepted into the distribution by Linus).
The first requires some explanation, because if I have already purchased
Minix, what posssible concern could price have for me? Simple.
If the OS is free, many more people will use/support/enhance it.
This is also the same reasoning I used when I bought my 386 instead
of a sparc (which I could have got for just 30% more). Since
PCs are cheap and generally available, more people will buy/use
them and thus good, cheap/free software will be abundant.
The second should be pretty obvious to anyone who has been using Minix
for for any period of time. AST generally does not accept enhancements
to Minix. This is not meant as a challenge, but merely a statement of
fact. AST has good and legitimate reasons for this, and I do not dispute
them. But Minix has some limitations which I just could no longer
live with, and due to this policy, the prospect of seeing them resolved
in reasonable time was unsatisfactory. These limitations include:
no 386 support
no virtual consoles
no soft links
no select call
no ptys
no demand paging/swapping/shared-text/shared-libs... (efficient mm)
chmem (inflexible mm)
no X-Windows (advocated for the same reasons as Linux and the 386).
no TCP/IP
no GNU/SysV integration (portability)
Some of these could be fixed by patches (and if you have done this
yourself, I don't have to tell you how satisfactory that is), but at
least the last 5 items were/are beyond any reasonable expectation.
Finally, my comment (crack?) about Minix's segmented kernel, or
micro-kernel architecture was more an expression of my frustration/
bewilderment at attempting to use the Minix PTY patches as a guide
of how to do it under Linux. That particular instance was one where
message passing greatly complicated the implementation of a feature.
I do have an opinion about Monlithic vs Message Passing, but won't
express it now, and did not mean to expresss it then. My goals are
totally short term (maximum functionality in the minimum amount of
time/cost/hassle), and so my views on this are irrelevant, and should
not be misconstrued. If you are non-plussed by the lack of the above
features, then you should consider Minix, as long as you don't mind
paying of course :)
From: olaf@oski.toppoint.de (Olaf Schlueter)
Subject: Re: Linux is obsolete
Date: 7 Feb 92 11:41:44 GMT
Organization: Toppoint Mailbox e.V.
Just a few comments to the discussion of Linux vs Minix, which evolved
partly to a discussion of monolithic vs micro-kernel.
I think there will be no aggreement between the two parties advocating
either concept, if they forget, that Linux and Minix have been designed
for different applications. If you want a cheap, powerful and
enhancable Unix system running on a single machine, with the possibility
to adapt standard Unix software without pain, then Linux is for you. If
you are interested in modern operating system concepts, and want to
learn how a microkernel based system works, then Minix is the better
choice.
It is not an argument against microkernel system, that for the time
being monolithic implemenations of Unix on PCs have a better
performance. This means only, that Unix is maybe better implemented as
a monolithic OS, at least as long as it runs on a single machine. From
the users point of view, the internal design of the OS doesn't matter at
all. Until it comes to networks. On the monolithic approach, a file
server will become a user process based on some hardware facility like
ethernet. Programs which want to use this facility will have to use
special libraries which offer the calls for communication with this
server. In a microkernel system it is possible to incorporate the
server into the OS without the need for new "system" calls. From the
users point of view this has the advantage, that nothing changes, he
just gets better performance (in terms of more disk space for example).
From the implementors point of view, the microkernel system is faster
adaptable to changes in hardware design.
It has been critized, that AST rejects any improvements to Minix. As he
is interested in the educational value of Minix, I understand his
argument, that he wants to keep the code simple, and don't want to
overload it with features. As an educational tool, Minix is written as
a microkernel system, although it is running on hardware platforms, who
will probably better perform with a monolithic OS. But the area of
network applications is growing and modern OS like Amoeba or Plan 9
cannot be written as monolithic systems. So Minix has been written with
the intention to give students a practical example of a microkernel OS,
to let them play with tasks and messages. It was not the idea to give a
lot of people a cheap, powerful OS for a tenth of the price of SYSV or
BSD implementations.
Resumee: Linux is not better than Minix, or the other way round. They
are different for good reasons.