 |
 | Home | Contact Us | FAQ | Search & Site Map | Link to Us |
 | Sign In | Join | Other 45 Sites in Network |
Mac Forum / Programming / CodeWarrior / December 2004Incrementation bug?
[[ This message was both posted and mailed: see the "To," "Cc," and "Newsgroups" headers for details. ]] The following snippet results in an infinite loop (as tested in CW 8.3 and CW 9.3, Mac). Specifically, a is not incremented. Can this be right? #include <stdio.h> int main(void) { int a = 0; int b = 10; while (a < b) { a = a++; } return 0; } Season's Greetings to all! Tony Patrinos www.escape.gr dans l'article 221220041128194651%patrinos@escape.gr, Anthony Patrinos à patrinos@escape.gr a écrit le 22/12/04 10:28 : > [[ This message was both posted and mailed: see > the "To," "Cc," and "Newsgroups" headers for details. ]] [quoted text clipped - 22 lines] > Tony Patrinos > www.escape.gr This is not a bug. The value copied to a is the value of a BEFORE a is incremented, and it is copied AFTER a is incremented. So the code you wrote reads as: while(a<b) { int x = a; a++; a = x; } Maybe you wanted to write: while(a<b) { a = ++a; } Eric > dans l'article 221220041128194651%patrinos@escape.gr, Anthony Patrinos à > patrinos@escape.gr a écrit le 22/12/04 10:28 : [quoted text clipped - 45 lines] > > Eric Thnx Eric for your reply... The assembly dump agrees with your interpretation of the correct order of evaluation. However, I wonder if the following order of evaluation is inconsistent with the language definition: 1. retrieve the old value of "a", 2. assign the old value of "a" to "a", 3. increment "a" At least this must be the way this works on some systems because I've encountered this situation while working on a port of the Panda PDF library (linux, windows) to MacOS X with CW. The code has been working for a number of years on the other systems. Researching further, I found a similar situation (a[i] = i++;) mentioned in K&R 2nd edition, p.53 as implementation dependent. Perhaps the compiler should issue a warning on situations susch as the above. Regards, Tony Patrinos > > dans l'article 221220041128194651%patrinos@escape.gr, Anthony Patrinos à > > patrinos@escape.gr a écrit le 22/12/04 10:28 : [quoted text clipped - 68 lines] > Perhaps the compiler should issue a warning on situations susch as the > above. Why is the assignment needed? Note that your original loop can be written: while (a < b) { ++a; } -- I> > Why is the assignment needed? Note that your original loop can be > written: [quoted text clipped - 3 lines] > > -- Well, it's not acually. But, as I mentioned, I came upon such a construct in the open source Panda PDF library. I compiled it for OSX, ran the sample program. It got stuck and debugging led to this "discovery"... -Tony Patrinos > dans l'article 221220041128194651%patrinos@escape.gr, Anthony Patrinos à > patrinos@escape.gr a écrit le 22/12/04 10:28 : [quoted text clipped - 45 lines] > > Eric Thnx Eric for your reply... The assembly dump agrees with your interpretation of the correct order of evaluation. However, I wonder if the following order of evaluation is inconsistent with the language definition: 1. retrieve the old value of "a", 2. assign the old value of "a" to "a", 3. increment "a" At least this must be the way this works on some systems because I've encountered this situation while working on a port of the Panda PDF library (linux, windows) to MacOS X with CW. The code has been working for a number of years on the other systems. Researching further, I found a similar situation (a[i] = i++;) mentioned in K&R 2nd edition, p.53 as implementation dependent. Regards, Tony Patrinos > [[ This message was both posted and mailed: see > the "To," "Cc," and "Newsgroups" headers for details. ]] [quoted text clipped - 22 lines] > Tony Patrinos > www.escape.gr I thought I'd test it with xCode (gcc back-end). The behavior is the opposite of CW and there is not infinite loop. -Tony Patrinos > [[ This message was both posted and mailed: see > the "To," "Cc," and "Newsgroups" headers for details. ]] [quoted text clipped - 22 lines] > Tony Patrinos > www.escape.gr The statement "a=a++;" has undefined behavior. C makes very few promises about the order of evaluation for expressions. Any expression that either modifies the same object twice, or modifies it and uses it elsewhere in the expression is undefined. So the following are all very bad... a = a++; x[a] = a++; y = a++ + a; The comp.lang.C FAQ (http://www.eskimo.com/~scs/C-faq/top.html) has some more detailed information about this (specifically questions 3.1, 3.2, and 3.9). Dave > [[ This message was both posted and mailed: see > the "To," "Cc," and "Newsgroups" headers for details. ]] [quoted text clipped - 22 lines] > Tony Patrinos > www.escape.gr I think this is undefined behavior according to the Standard, beacause the value is written more than once between two sequence points. Gianluca |
Reply to this Message |
 Sign In
 Join
 My Latest Posts My Monitored Threads My Blog My Photo Gallery My Profile My Homepage Start New Thread Enable EMail Alerts Rate this Thread
|
©2008 Advenet LLC Privacy Policy - Terms of Use
This website includes both content owned or controlled by Advenet as well as content owned or controlled by third parties.