Early C ?

Main differences that I can think of:

(1) no 'void' in those days.

(2) function declaration is now rather different

(3) Initializer syntax is different ("int x = 1;" rather than "int x 1;")

(4) =operator replaced by operator= ("x += y;" rather then "x =+ y;")

(5) I think short and long modifiers weren't there at the time either; you got 'int' and 'int' was the size of a machine word.

(6) I'm not sure C had unions back in those days, which made certain types of systems programming task harder. (Exotica like bitfields were certainly not there, too).

pete

Thanks, Pete

very helpful

martin

(clarifying context restored)

... snip ...

(7) structs could not be passed or returned by value.

The horse's mouth is of course Dennis Ritchie, you can find more discussion of early C including documents and source code from the period here:

I should have pointed out that there are still remnants of this around today. The single struct namespace is why a lot of the system structures have redundant looking prefixes on the field names:

For example:

struct timeval { long tv_sec; /* seconds */ long tv_usec; /* microseconds */ }; Then in user code: struct timeval tv;

tv.tv_sec = 1; tv.tv_usec = 0; Why the tv_ on the front of the field names? This seems cleaner:

tv.sec = 1; tv.usec = 0;

But, that would have caused problems if some other structure definition used "sec" or "usec" as field names. So you put a prefix on all structure field names to try to avoid collisions.

Thanks. I came to the C party a little later ('86), and started working with pretty much the "K&R first edition" version of the language - so I was working backwards from what I remember of dmr's writings and the examples in that tutorial!

pete

But in K&R, passing, returning and assigning structures wasn't allowed, this was new in K&R2 ('88).