Why C for the Open Source Movement?

Why C for the Open Source Movement?

[zeta_no]

Hi to all,

Can someone explain me the reasons why the main actors of the Open
Source community didn't choose Ada instead of C to write the core
elements of their systems (Linux, Hurd, FreeBSD etc...) and most of the
critical applications that come with it if the main goal was to provide
a real alternative to the "proprietary clan".  Would it have reduce
significantly bug list for all devel. projects and assure much more
reliable applications?  Isn't the Unix tradition based on well crafted
design mecanisms?  Does its traditional ways became so heavy to prevent
adaptation and or renewal of its philosophy through time?

Hope I do not make a troll of me,

olivier h.

Re: Why C for the Open Source Movement?

[Jeffrey Creem]

zeta_no wrote:
> Hi to all,
> 
> Can someone explain me the reasons why the main actors of the Open
> Source community didn't choose Ada instead of C to write the core
> elements of their systems (Linux, Hurd, FreeBSD etc...) and most of the
> critical applications that come with it if the main goal was to provide
> a real alternative to the "proprietary clan".  Would it have reduce
> significantly bug list for all devel. projects and assure much more
> reliable applications?  Isn't the Unix tradition based on well crafted
> design mecanisms?  Does its traditional ways became so heavy to prevent
> adaptation and or renewal of its philosophy through time?
> 
> Hope I do not make a troll of me,
> 
> olivier h.
> 

All of those projects started before there was a freely available Ada 
compiler.

In the case of someting like the GNU project itself, when Stallman 
started the GNU project, Ada 83 was not even really out yet and 
certainly not in a position where it was stable.


There is a lot of inertia to overcome once one starts an OS project. 
Switching the Linux kernel or Hurd or anything else of that scale to Ada 
just for the sake of the better language is probably not the best idea.

Ignoring all of the forking and grubling that it would create for a 
moment at the very least one would have to agree that it could be years 
before you got back to the same level of stability as the original kernel.

One could argue that it could be done piecewise but the problem with 
that is if you start small now you have to justify pulling in some small 
Ada runtime into the kernel just to support some particular module and 
again it does not seem to make sense.

There are lots of "silver bullet" languages out there that have cool 
features (e.g. Python) but still no one has tried to port the kernel to 
python.

Re: Why C for the Open Source Movement?

[kevin cline]

Jeffrey Creem wrote:
> zeta_no wrote:
> > Hi to all,
> >
> > Can someone explain me the reasons why the main actors of the Open
> > Source community didn't choose Ada instead of C to write the core
> > elements of their systems (Linux, Hurd, FreeBSD etc...) ,,,
>
> All of those projects started before there was a freely available Ada
> compiler.

And even if there had been a compiler, Ada-83 was ill-suited for the
work.  The language was crippled by design to prevent certain classes
of bugs.  Unfortunately, this also made certain classes of applications
difficult to develop.  Nor does Ada's verbosity seem to appeal to the
majority of programmers motivated to do public-domain work.

Re: Why C for the Open Source Movement?

[Ludovic Brenta]

Jeffrey Creem writes :
> All of those projects started before there was a freely available Ada
> compiler.
>
> In the case of someting like the GNU project itself, when Stallman
> started the GNU project, Ada 83 was not even really out yet and
> certainly not in a position where it was stable.
[...]

I think C was well suited as a bootstrap language, because C was
designed to make the compiler writer's job easy. When starting the GNU
project, RMS first wrote a C compiler, and then used that to write a
Lisp system, which he used for Emacs. That was the right thing to do,
IMHO.

Early on, Lisp was therefore in a much better position than Ada to
become the language of choice for GNU. Even today, there is still quite
a lot of Lisp in GNU/Linux systems; and not just in Emacs and XEmacs,
as there are several free Common Lisp and Scheme systems available now.
But even then, Lisp is still a minority language.  So the OP's question
remains valid, but in the context of the early days of the GNU project,
I would ask "why C and not Lisp"? I think it may be because Lisp has or
can be perceived to have too much overhead for writing libraries or a
kernel, and so was summarily and incorrectly dismissed by application
programmers. I would think the folks on comp.lang.lisp have already
discussed this question many times.

-- 
Ludovic Brenta.

Re: Why C for the Open Source Movement?

[Ludovic Brenta]

zeta_no writes :
> Isn't the Unix tradition based on well crafted design mecanisms?

No, it's not. It started as a hack made by a few long-haired, rebel
programmers in their spare time, bazaar-style. In contrast, MULTICS was
intended to be a beautiful cathedral of software, and I think it was
written in PL/I not C. MULTICS is still not complete, and I think it'll
take as long to complete as it took actual cathedrals :)

But history showed that "worse is better", unfortunately for purists.

http://en.wikipedia.org/wiki/Worse_is_better

-- 
Ludovic Brenta.

Re: Why C for the Open Source Movement?

[Georg Bauhaus]

Ludovic Brenta wrote:
> zeta_no writes :
>> Isn't the Unix tradition based on well crafted design mecanisms?
> 
> No, it's not. It started as a hack made by a few long-haired, rebel
> programmers in their spare time, bazaar-style.

This contrasts with what I have read in the past.

> MULTICS is still not complete, and I think it'll
> take as long to complete as it took actual cathedrals :)


Reportedly, there were a number if quite complete installations,
even though some features were changed, or left out. The project
was even saved from cancellation by someone who became aware what
had been achieved.

http://www.multicians.org/history.html 

Re: Why C for the Open Source Movement?

[Dmitry A. Kazakov]

On 8 May 2006 02:13:22 -0700, Ludovic Brenta wrote:

> zeta_no writes :
>> Isn't the Unix tradition based on well crafted design mecanisms?
> 
> No, it's not. It started as a hack made by a few long-haired, rebel
> programmers in their spare time, bazaar-style. In contrast, MULTICS was
> intended to be a beautiful cathedral of software, and I think it was
> written in PL/I not C. MULTICS is still not complete, and I think it'll
> take as long to complete as it took actual cathedrals :)
> 
> But history showed that "worse is better", unfortunately for purists.
> 
> http://en.wikipedia.org/wiki/Worse_is_better

I don't think that it was simplicity. UNIX never was simple, rather it was
simply bad. (:-))

P.S. It is illogical to express badness in a positive way, like
"simplicity", for example. If simplicity is the goal, then it is good, and
thus bad cannot be simple! In fact, to be really bad (as UNIX, or Windows
is) is much harder than to be any good. So many things can be considered
positive... It is very difficult to ensure that most combinations of them
were indeed wrong... (:-))

-- 
Regards,
Dmitry A. Kazakov
http://www.dmitry-kazakov.de

Re: Why C for the Open Source Movement?

[thvv]

Ludovic Brenta wrote:
> zeta_no writes :
> > Isn't the Unix tradition based on well crafted design mecanisms?
>
> No, it's not. It started as a hack made by a few long-haired, rebel
> programmers in their spare time, bazaar-style. In contrast, MULTICS was
> intended to be a beautiful cathedral of software, and I think it was
> written in PL/I not C. MULTICS is still not complete, and I think it'll
> take as long to complete as it took actual cathedrals :)

Dick Gabriel is welcome to his opinion, but as someone who worked
on Multics with the creators of UNIX in the 60s I have to say that I
don't
see it that way. Multics was not rigorously planned first and then
executed; our design and implementation evolved over 20 years.
We did have this pattern of writing something and discussing it before
coding, and often the discussion led to an improved design. Multicians
felt that we were rebels, back in the 60s, fighting against batch
processing,
writing systems in assembler, and using human waves of programmers.
We advocated flexibility, virtual memory, interactiveness, and powerful
tools.  Ken and Dennis did not have the longest hair or wildest beards
on the Multics development team.

Multics was written in PL/I. The Bell Labs folks encountered BCPL
at Project MAC in the 60s and created B and then C in the 70s.
Multics was not "complete" when Bull stopped development on it in
the mid 80s: no operating system is ever "complete" since user needs,
hardware, and competitiion keep changing requirements.

For more information, see
  http://www.multicians.org/myths.html
  http://www.multicians.org/pl1.html

Re: Why C for the Open Source Movement?

[Ludovic Brenta]

thvv writes :
> Ludovic Brenta wrote:
>> zeta_no writes :
>>> Isn't the Unix tradition based on well crafted design mecanisms?
>>
>> No, it's not. It started as a hack made by a few long-haired, rebel
>> programmers in their spare time, bazaar-style. In contrast, MULTICS was
>> intended to be a beautiful cathedral of software, and I think it was
>> written in PL/I not C. MULTICS is still not complete, and I think it'll
>> take as long to complete as it took actual cathedrals :)
>
> Dick Gabriel is welcome to his opinion,

Who is Dick Gabriel?

> but as someone who worked on Multics with the creators of UNIX in
> the 60s I have to say that I don't see it that way. Multics was not
> rigorously planned first and then executed; our design and
> implementation evolved over 20 years.  We did have this pattern of
> writing something and discussing it before coding, and often the
> discussion led to an improved design. Multicians felt that we were
> rebels, back in the 60s, fighting against batch processing, writing
> systems in assembler, and using human waves of programmers.  We
> advocated flexibility, virtual memory, interactiveness, and powerful
> tools.  Ken and Dennis did not have the longest hair or wildest
> beards on the Multics development team.

:-) Thanks a lot for correcting me.  I am honoured that you gave us
this first-hand information.  Multics is quite a legend to me, a bit
like dragons; now I feel like I've just met an actual dragon slayer :)

Still, would you agree that Multics used The Right Thing approach, as
opposed to Unix which uses Worse Is Better?

-- 
Ludovic Brenta.

Re: Why C for the Open Source Movement?

[Ed Falis]

Ludovic Brenta wrote:

> Who is Dick Gabriel?

> Still, would you agree that Multics used The Right Thing approach, as
> opposed to Unix which uses Worse Is Better?
>

The guy who wrote "Worse is Better" ;-)

- Ed