Article View: comp.lang.modula2
Article #2433Re: c and m2
From: Lawrence C Smith
Date: Tue, 26 Jan 1993 18:12
Date: Tue, 26 Jan 1993 18:12
42 lines
2597 bytes
2597 bytes
In article <1993Jan25.184552.23393@mcshub.dcss.mcmaster.ca>, cs4gp6au@maccs.mcmaster.ca (Lockie RJ) writes: >/* p now points to test[1] according to K&R2 pg. 98 */ > (* p now points to test[0][1] *) >So, if the above code is correct, the question I have is why does >incrementing a pointer in C point to the next element in the array? >while in Modula-2 it doesn't (the pointer needs to be incremented >by the size of the array element). >How does C do that? Is it the compiler? In Modula-2, to increment to >the next element in the array it would require: > INC( p, SIZE(ex) ); The main difference between B and C is the notion of types. B just had words, C had ints, and chars, and so on. One of the original reasons for this change was so that the compiler could recognize pointer to <type> and increment by the appropriate number of bytes to get to the next element. The reason this is so is because arrays are second-class citizens in C, merely vectors of elements, not a real type, like structs are. You can see this in every function call, where an array is always passed by address (even the name is merely a constant referring to the address of the first element) and a struct is actually copied onto the stack, like any other parameter. Vectors were set up so the programmer could scan them by walking a pointer up and down - in fact, C is completely designed around this assembly-code concept. Modula and Pascal promote arrays to first-class citizens, they are treated just like records and they define _subscripting_ as the access mechanism, _not_ pointer dereferencing. When you inc or dec a pointer you are poten- tially breaking typechecking by potentially breaking range-checking. Modula's INC builtin is defined to increment its argument by one, no more. It is _not_ defined as a means of stepping to the next "reasonable" value. It is _not_ a pointer operation. Larry Smith (smith@ctron.com) No, I don't speak for Cabletron. Need you ask? - Liberty is not the freedom to do whatever we want, it is the freedom to do whatever we are able.
Technical Details
Message-ID:
<6568@balrog.ctron.com>Path:
arc1-ca.pugleaf.net!sparky!uunet!charon.amdahl.com!pacbell.com!sgiblab!spool.mu.edu!samsung!balrog!ctron.comReferences:
<1993Jan25.184552.23393@mcshub.dcss.mcmaster.ca>Show Full Headers
Complete Article Headers
Path: arc1-ca.pugleaf.net!sparky!uunet!charon.amdahl.com!pacbell.com!sgiblab!spool.mu.edu!samsung!balrog!ctron.com From: smith@ctron.com (Lawrence C Smith) Newsgroups: comp.lang.modula2 Subject: Re: c and m2 Keywords: c, m2 Message-ID: <6568@balrog.ctron.com> Date: 26 Jan 93 18:12:45 GMT References: <1993Jan25.184552.23393@mcshub.dcss.mcmaster.ca> Sender: usenet@balrog.ctron.com Reply-To: smith@ctron.com Organization: Cabletron Systems, Inc. Lines: 42 Nntp-Posting-Host: glinda