🚀 go-pugleaf

In article <75055@cc.utah.edu> RCAPENER@cc.utah.edu writes:

>In article <7734@tekgvs.LABS.TEK.COM>, toma@tekgvs.LABS.TEK.COM
>(Tom Almy) writes:
>> I bought the upgrade, and then read this group. While I'm not happy with
>> the C orientation of V2.00 (mainly because the 8086 is a lousy machine for
>> a C implementation, and much better for Modula-2 or Pascal which can
>> effectively handle segments) the complaint about missing "C language sources"
>> is blatently false.
>>
>
>Jimminy Crickets!  Is that why they use C to write almost all of the
>software for IBM RS/6000, the Motorola 88000 based machines (DG Aviion
>and Tektronix WS)?  Hmm, I wonder why C works so well on those machines
>that have segmented architectures.  Perhaps it could be that they are
>workstations while the lowly 8086 is only used in a PC 8^)?

The reasons for choosing C as the implementation language are often entirly
divorced from the architecture of the machine.  Where unix is the OS there
is no choice, other languages are either not available or poorly implemented /
supported (eg Logitech stopped selling and supporting MODULA-2 for Xenix).
This may be changing, but as of today C is the best bet for unix machines.
With other OS's it depends on your OS / machine;  in the MSDOS world
Modula-2 is an excellent choice (IMHO), on a CDC 6600 FORTRAN and so on.
The choice of language must, in part, be guided by the level of support
for the language. Also, in all too many cases, C is chosen because
'everyone else uses C', 

Tom Almy's point about the suitablity of C for a segmented architecture like
that of 8086 (and 80x86's in real mode) family is entirely valid.  First,
it is important to stress that the following is only applicable when
attempting to write large applications on such an architecture (in this context
large is defined as a code or data size > segment size).  C does not
encourage modularization of code in the same way Modula-2 does, in particular
a single Modula-2 module can be compiled into one segment (assuming it's 
small enough) and only the functions that are exported and imported need
be called using far calls.  Likewise with the data, data is local unless
exported, each module can have it's own data segment.  Now C people
would say, "but we can do that too, just stick static infront of all the
local functions and variables".  The problem is that this is not the
default and in general, usually because of programmer laziness, all
functions end up being global.

To summarize, Modula-2's Modules, are better suited to compilation into
seperate segments than C's compilation units (files).

>I delight in pulling the tiger's tail 8^))).
>
Careful, the head may turn round and bite :-)

>Or maybe the fact that the majority of the compilers out there are
>written in C (ok, so some of them are still written in Assembler)

Most compilers out there are not written in C, most C compilers are written
in C (not all - I beleive Topspeed C is written in Modula-2).  A few
of the older compilers are written in assembler (most notably Borlands
TurboPascal for CP/M (remember CP/M!)).  The rest are written in a very
mixed bag of languages, depending on there age and the manufacturer.

C, despite its age, is a relatively new language.  Very little programming,
outside unix, was done in C even 10 years ago. Now every man and his dog
has a C compile

Steve
-- 
Steve.Bleazard@RoboBar.Co.Uk        | Phone:  +44 81 991 1142 x153
Snr Software Engineer, Robobar Ltd. | Fax:    +44 81 998 8343 (G3)
22 Wadsworth Road, Perivale.        |
Middx., UB6 7JD ENGLAND.            | ...!ukc!robobar!steve