unexpected result

[Jim Hester]

The following stemmed from a slightly bizarre line in a student's work that
surprised my by doing what HE wanted it to do. Question: Are the second and
third cases below correct by some interpretation of the language semantics
that I'm missing, or is it some kind of register-optimization bug (or
something else)?

int i = 1;

int j;

j = i++; // i ends up 2, j ends up 1 (as expected)

cout << "i=1; j = i++; i == " << i << " j == " << j << endl << endl;



int k = 1;

k = k++; // k ends up 2 - not exptectd.

cout << "K=1; k = k++; k == " << k << endl << endl;



int m = 1;

int p;

m = ( p = m++ ); // p ends up 1 as expected, but m remains 2 (not set back
to 1)

cout << "m=1; m = (p = m++); p == " << p << " m == " << m << endl << endl;

 

[Victor Bazarov]

The following stemmed from a slightly bizarre line in a student's work that
surprised my by doing what HE wanted it to do. Question: Are the second and
third cases below correct by some interpretation of the language semantics
that I'm missing, or is it some kind of register-optimization bug (or
something else)?

[... i=i++ again ...]

Search in the news archives for "undefined behaviour".

V

 

[Default User]

Jim Hester wrote:

int k = 1;

k = k++; // k ends up 2 - not exptectd.

cout << "K=1; k = k++; k == " << k << endl << endl;

Not sure if the C++ FAQ covers it, but it's the same as in C:

http://c-faq.com/expr/ieqiplusplus.html


Brian

 

[red floyd]

k = k++; // k ends up 2 - not exptectd.

m = ( p = m++ ); // p ends up 1 as expected, but m remains 2 (not set back
to 1)

Both of these invoke undefined behavior. It is free to work "as
expected", or do anything else, including blow up your monitor, reformat
your hard drive, or even contact the Pentagon and start WWIII.

 

[Ben Pope]

The following stemmed from a slightly bizarre line in a student's work that
surprised my by doing what HE wanted it to do. Question: Are the second and
third cases below correct by some interpretation of the language semantics
that I'm missing, or is it some kind of register-optimization bug (or
something else)?

int i = 1;

int j;

j = i++; // i ends up 2, j ends up 1 (as expected)

cout << "i=1; j = i++; i == " << i << " j == " << j << endl << endl;



int k = 1;

k = k++; // k ends up 2 - not exptectd.

This is undefined behaviour. Search for "sequence point".

Ben Pope