28 Aug
2006
28 Aug
'06
5:14 p.m.
Subject: Re: "File types"
From: "Ethan Dicks" <ethan.dicks at gmail.com>
Date: Tue, 29 Aug 2006 11:52:56 +1200
To: "General Discussion: On-Topic and Off-Topic Posts" <cctalk at
classiccmp.org>
On 8/29/06, Chuck Guzis <cclist at sydex.com> wrote:
> Unix was the first system that I'd ever seen where the file name implied a
> type.
For me it was OS8 and later tops 10. By RT11 and CP/M the idea that
COM, .TXT, .DOC and .ASM were fully embedded.
And even CP/M apps were somewhat sensitized... Ever type A:ASM FOO.ASM and
wonder why you got a disk specification error?
Allison
28 Aug
28 Aug
11:27 p.m.
On 8/28/2006 at 8:14 PM Allison wrote:
And even CP/M apps were somewhat sensitized... Ever
type A:ASM FOO.ASM and
wonder why you got a disk specification error?
Yeah, but ASM was just trying to get away with using a single FCB to
specify all three files. So you get A:ASM FOO.AAZ or some such to do what
you want. ASM is an aberration.
That's another thing--command syntax that's definitely from the mini/micro
convention. Up until CP/M, I'd always done something like ASM IN=FOO
OBJECT=FOO2 LIST=FOO3
I really DON'T like switch-type command-line structure.
OT tidbit: With CDC 6000 FORTRAN the idea was that the compiler would
read the first statement of the inpjut then determine what needed to be
done. Most newbies were puzzled by it being called RUN. But you could
stack COMPASS assembly or FORTRAN source in your input deck and simply say
"RUN." AFAIK, it was never extended to other languages, but it was pretty
neat.
Cheers,
Chuck
29 Aug
29 Aug
4:10 p.m.
And even CP/M apps were somewhat sensitized... Ever
type A:ASM FOO.ASM and
wonder why you got a disk specification error?
Doesn't it try to write the object to S:, and the listing to M: ? Neither
of which probably exist.
I remember the first time I used ZASM (a CP/M Z80 assembler) on an
RML380Z. I typed 'ZASM FOO,ZSM and got a 'BDOS ERROR ON Z: SELECT' for my
efforts....
To return to the original question, how far back do you have to go for
unix's cc(1) to require .c on C source files, .o on object files (to be
linked it), and so on?
-tony
3:48 p.m.
It was thus said that the Great Tony Duell once stated:
Not very far.
[spc]linus:~/source/cgi/src>gcc --version
2.7.2.3
[spc]linus:~/source/cgi/src>cp pair.c pair.blah
[spc]linus:~/source/cgi/src>gcc -c pair.blah
gcc: pair.blah: linker input file unused since linking not done
[spc]linus:~/source/cgi/src>gcc -c pair.c
[spc]linus:~/source/cgi/src>ls pair.*
pair.blah pair.c pair.h pair.o
[spc]linus:~/source/cgi/src>
That's late 1997 (still running a RedHat 5.2 system). Checking a newer
system (install just two weeks ago):
[root at spodie tmp]# gcc -c hello.blah
gcc: hello.blah: linker input file unused because linking not done
[root at spodie tmp]# gcc -c hello.c
[root at spodie tmp]# ls hello.*
hello.blah hello.c hello.o
So it looks like GCC doesn't like arbitrary extentions.
-spc (which actually surprised me ... )
And even CP/M
apps were somewhat sensitized... Ever type A:ASM FOO.ASM and
wonder why you got a disk specification error?
Doesn't it try to write the object to S:, and the listing to M: ? Neither
of which probably exist.
I remember the first time I used ZASM (a CP/M Z80 assembler) on an
RML380Z. I typed 'ZASM FOO,ZSM and got a 'BDOS ERROR ON Z: SELECT' for my
efforts....
To return to the original question, how far back do you have to go for
unix's cc(1) to require .c on C source files, .o on object files (to be
linked it), and so on? 
6:45 p.m.
On Tue, 29 Aug 2006, Sean Conner wrote:
> To return to the original question, how far back
do you have to go for
> unix's cc(1) to require .c on C source files, .o on object files (to be
> linked it), and so on?
[snip tests]
So it looks like GCC doesn't like arbitrary
extentions.
This is because gcc is a compiler and linker _driver_. It needs to know if
the input file is source in c, c++, objective-c, assembler, ada, f77, f95,
java, treelang, or is an object file in a.out, elf, coff, etc. format so
it can invoke the right backend. You can tell gcc what to do explicitly
with the -x option:
$ ls
test.foo
$ gcc -x c -c test.foo
$ ls
test.foo test.o
$ rm test.o
$ ls
test.foo
$ gcc -x c test.foo
$ ls
a.out test.foo
Remember, gcc stands for "GNU Compiler Collection", not "GNU C
Compiler".
Alexey
10:48 p.m.
> To return to the original question, how far back
do you have to go
> for unix's cc(1) to require .c on C source files, .o on object files
> (to be linked it), and so on?
Well, cc is usually a wrapper; if you break it down and run the stages
manually, I don't know of any version which enforces any extensions.
Not very far. [...gcc example...]
gcc (a) is a driver wrapper and (b) has options (which you conveniently
didn't mention) to explicitly specify what kind of input and output you
want it to work with, overriding the defaults (which are based on file
names).
/~\ The ASCII der Mouse
\ / Ribbon Campaign
X Against HTML mouse at rodents.montreal.qc.ca
/ \ Email! 7D C8 61 52 5D E7 2D 39 4E F1 31 3E E8 B3 27 4B
8:29 p.m.
It was thus said that the Great der Mouse once stated:
I actually didn't realize that when I did the example. I guess that's
what I get for sticking with C for the most part.
-spc (Still learning stuff about Unix even after 16 years ... )
Not very far. [...gcc example...]
gcc (a) is a driver wrapper and (b) has options (which you conveniently
didn't mention) to explicitly specify what kind of input and output you
want it to work with, overriding the defaults (which are based on file
names). 
30 Aug
30 Aug
1:30 a.m.
Hi,
Tony Duell said:
They're present in 5th Edition (June 1974) but whether they're _required_
or not I can't say, I don't have the machine switched on atm.
..
..
OK I booted v5 up and trying to "cc hello" after renaming hello.c
to hello results in "Format error: hello".
Just a quick test...
--
Cheers,
Stan Barr stanb at dial.pipex.com
The future was never like this!
And even CP/M
apps were somewhat sensitized... Ever type A:ASM FOO.ASM and
wonder why you got a disk specification error?
Doesn't it try to write the object to S:, and the listing to M: ? Neither
of which probably exist.
I remember the first time I used ZASM (a CP/M Z80 assembler) on an
RML380Z. I typed 'ZASM FOO,ZSM and got a 'BDOS ERROR ON Z: SELECT' for my
efforts....
To return to the original question, how far back do you have to go for
unix's cc(1) to require .c on C source files, .o on object files (to be
linked it), and so on?
6688
days inactive
6689
days old
7 comments
7 participants
participants (7)
-
ajp166＠bellatlantic.net -
alexeyt＠freeshell.org -
ard＠p850ug1.demon.co.uk -
cclist＠sydex.com -
mouse＠Rodents.Montreal.QC.CA -
spc＠conman.org -
stanb＠dial.pipex.com