C++Talk.NET C++ language newsgroups   Archives   FAQ   Search   Memberlist   Usergroups   Register   Profile   Log in to check your private messages   Log in  sequence point and reference          C++Talk.NET Forum Index -> C++ Language (Moderated) View previous topic :: View next topic   Author Message Lee Guest Posted: Sat Feb 28, 2004 12:22 pm    Post subject: sequence point and reference Dear moderator, please delete my previous post, I made a stupid mistake, please accept my apology. The typo confues everybody. {We have nine moderators, nonetheless both your posts arrived at me but lacking a time machine... If you want to kill an earlier post always email the moderators with the tracking number with a request for it to be rejected. I am adding this comment because others sometimes face the same problem. -mod} My question should be as follows: ********************************************************************** Hello, everybody, I have a midterm question as follows, which I believe is UB. I think the statement j*f(i,j) is similar to j*(--j), which leads to undefined behaviour (UB). But our professor insists the only correct result is 4. Any comments? int f(int i, int& j) { if(j==1) return i; --j; ++i; return j*f(i, j); } int main() { int n=4; cout< return 0; } Thanks, Lee [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Dhruv Matani Guest Posted: Sun Feb 29, 2004 2:56 am    Post subject: Re: sequence point and reference On Sat, 28 Feb 2004 07:22:34 -0500, Lee wrote: Quote: I have a midterm question as follows, which I believe is UB. I think the statement j*f(i,j) is similar to j*(--j), which leads to undefined behaviour (UB). But our professor insists the only correct result is 4. Any comments? Looking at the code below, I can almost hear the UB coming out! Yes. Even I think that the code below produces results that are not defined correctly. The value of 'j' may be cached before each call to f() and also it may not be (as your professor is assuming it is). Quote: int f(int i, int& j) { if(j==1) return i; --j; ++i; return j*f(i, j); } int main() { int n=4; cout< return 0; } I don't know what cout<< will print after all the functions return. I believe that this produces multiple UBs! -- Regards, -Dhruv. Proud to be a Vegetarian. http://www.vegetarianstarterkit.com/ http://www.vegkids.com/vegkids/index.html [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Roger Orr Guest Posted: Sun Feb 29, 2004 2:58 am    Post subject: Re: sequence point and reference Hi Lee, "return j * f(i,j)" is similar to "return j * (--j )", but different enough to remove the UB. The difference is that, when calling a function, "there is also a sequence point after the copying of a returned value and before the execution of any expressions outside the function" (1.9p17 in the standard) This means that the code is executed a bit like this: int seqpt = f(i,j); return j*seqpt; The presence of the additional sequence point removes the ambiguity. Roger Orr -- MVP in C++ at www.brainbench.com [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top John Potter Guest Posted: Sun Feb 29, 2004 3:33 am    Post subject: Re: sequence point and reference On 28 Feb 2004 07:22:34 -0500, [email]graceleeok (AT) yahoo (DOT) com[/email] (Lee) wrote: Quote: I have a midterm question as follows, which I believe is UB. I think the statement j*f(i,j) is similar to j*(--j), which leads to undefined behaviour (UB). The second is undefined because it modifies j and uses j without a sequence point. The first does not have that problem because there is a sequence point between evaluating the parameters to f and calling f and another one between setting the return value of f and using it in the caller. There is no undefined behavior in the example. Quote: But our professor insists the only correct result is 4. Any comments? Never trust a professor, including me. :) Quote: int f(int i, int& j) { if(j==1) return i; --j; ++i; return j*f(i, j); This is the point of contention. There is an lvalue to rvalue conversion on i to send that parameter which is no problem. There is no conversion on j to send the reference which is no problem. There is an lvalue to rvalue conversion on j which is the problem. The order in which the subexpressions of the multiplication are evaluated is unspecified. The lvalue to rvalue conversion on j may be performed before or after the call of f. There is not even any requirement that it be done in the same order each time that the function is called. The following valid translation of that code is interesting. int k(j); int r(f(i, j)); return (rand() % 2 ? j : k) * r; The possible results are 4, 8, 12, 24. Note that this is unspecified behavior not undefined behavior. There are no other possible results. With undefined behavior, anything can happen. John [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Francis Glassborow Guest Posted: Sun Feb 29, 2004 3:36 am    Post subject: Re: sequence point and reference In message <e80c8ec3.0402271915.2bfc5691 (AT) posting (DOT) google.com>, Lee <graceleeok (AT) yahoo (DOT) com> writes Quote: Dear moderator, please delete my previous post, I made a stupid mistake, please accept my apology. The typo confues everybody. {We have nine moderators, nonetheless both your posts arrived at me but lacking a time machine... If you want to kill an earlier post always email the moderators with the tracking number with a request for it to be rejected. I am adding this comment because others sometimes face the same problem. -mod} My question should be as follows: ********************************************************************** Hello, everybody, I have a midterm question as follows, which I believe is UB. I think the statement j*f(i,j) is similar to j*(--j), which leads to undefined behaviour (UB). But our professor insists the only correct result is 4. Any comments? Your professor is right as regards UB (well apart from the horrible code, but I guess that was to make a point). Nowhere is j written to so the rules about reading in order to modify are not operative. The change of j is safely encapsulated in a function and so protected by sequence points. I think the result is actually unspecified because the compiler is free to evaluate f(i, j) either before or after evaluating j (and is allowed but not required to re-read j for the two occurrences. Quote: int f(int i, int& j) It might have been more consistent to have passed i by reference as well Quote: { if(j==1) return i; --j; ++i; return j*f(i, j); } int main() { int n=4; cout< return 0; } f(1, 4) calls f(2, 3) calls f(3, 2) calls f(4, 1). That last returns 4 but what value is used for the first j in j*f(i, j) depends on when the compiler evaluates it. If, as it is entitled to do, it evaluates it before calling the function you will get a different result from that which you get if it evaluates it after the function returns. This is not undefined behaviour (because the changes to j take place properly protected by sequence points but it is unspecified behaviour because C++ (as well as C but not Java) does not lay down the order of evaluation. For example, f(1, 4) = 4*f(2,3) = 4*3*f(3,2) = 4*3*2*f(4, 1) = 4*3*2*4 is, I think, perfectly consistent with the rules for evaluation in C++. -- Francis Glassborow ACCU Author of 'You Can Do It!' see http://www.spellen.org/youcandoit For project ideas and contributions: http://www.spellen.org/youcandoit/projects [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Francis Glassborow Guest Posted: Sun Feb 29, 2004 4:43 pm    Post subject: Re: sequence point and reference In message <pan.2004.02.28.14.26.52.481923 (AT) gmx (DOT) net>, Dhruv Matani <dhruvbird (AT) gmx (DOT) net> writes Quote: Looking at the code below, I can almost hear the UB coming out! Yes. Even I think that the code below produces results that are not defined correctly. That is not the same as undefined. The code has unspecified behaviour, but the compiler is not allowed to replace it with an implementation of nethack. :-) Quote: The value of 'j' may be cached before each call to f() and also it may not be (as your professor is assuming it is). Yes, that is why the result is unspecified. Quote: int f(int i, int& j) { if(j==1) return i; --j; ++i; return j*f(i, j); } int main() { int n=4; cout< return 0; } I don't know what cout<< will print after all the functions return. I believe that this produces multiple UBs! Then you believe wrong. There is no undefined behaviour in the above code. -- Francis Glassborow ACCU Author of 'You Can Do It!' see http://www.spellen.org/youcandoit For project ideas and contributions: http://www.spellen.org/youcandoit/projects [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Carsten Hansen Guest Posted: Sun Feb 29, 2004 5:33 pm    Post subject: Re: sequence point and reference "Lee" <graceleeok (AT) yahoo (DOT) com> wrote Quote: Dear moderator, please delete my previous post, I made a stupid mistake, please accept my apology. The typo confues everybody. {We have nine moderators, nonetheless both your posts arrived at me but lacking a time machine... If you want to kill an earlier post always email the moderators with the tracking number with a request for it to be rejected. I am adding this comment because others sometimes face the same problem. -mod} My question should be as follows: ********************************************************************** Hello, everybody, I have a midterm question as follows, which I believe is UB. I think the statement j*f(i,j) is similar to j*(--j), which leads to undefined behaviour (UB). But our professor insists the only correct result is 4. Any comments? int f(int i, int& j) { if(j==1) return i; --j; ++i; return j*f(i, j); } int main() { int n=4; cout< return 0; } Thanks, Lee The order in which the two terms in the expression j * f(i,j) are evaluated is unspecified. Hence the result of the calculation cannot be determined. Carsten Hansen [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Dhruv Matani Guest Posted: Mon Mar 01, 2004 5:11 pm    Post subject: Re: sequence point and reference On Sun, 29 Feb 2004 11:43:55 -0500, Francis Glassborow wrote: Quote: In message <pan.2004.02.28.14.26.52.481923 (AT) gmx (DOT) net>, Dhruv Matani [email]dhruvbird (AT) gmx (DOT) net[/email]> writes Looking at the code below, I can almost hear the UB coming out! Yes. Even I think that the code below produces results that are not defined correctly. That is not the same as undefined. The code has unspecified behaviour, but the compiler is not allowed to replace it with an implementation of nethack. Sorry! I agree I'm wrong here. I remember having this discussion about the difference between UB and Defined B with Unspecified B. However, form a programmer's point of view, I would only care about whether the following would: 1. Compile. 2. Run. 3. Produce the same results always on different platforms or even different versions of the same compiler or even the same compiler! (3) Basically means not doing anything that is explicitly specified by the standard to be UB or if the standard says that something is implementation defined then I should try and not get tied by some particular implementation. Quote: The value of 'j' may be cached before each call to f() and also it may not be (as your professor is assuming it is). Yes, that is why the result is unspecified. Ok. Quote: I don't know what cout<< will print after all the functions return. I believe that this produces multiple UBs! Then you believe wrong. There is no undefined behaviour in the above code. Then would multiple Unspecified Behaviours be the correct term to use? -- Regards, -Dhruv. Proud to be a Vegetarian. http://www.vegetarianstarterkit.com/ http://www.vegkids.com/vegkids/index.html [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Display posts from previous: All Posts1 Day7 Days2 Weeks1 Month3 Months6 Months1 Year Oldest FirstNewest First         C++Talk.NET Forum Index -> C++ Language (Moderated) All times are GMT Page 1 of 1   Jump to: Select a forum C/C++----------------C LanguageC Language (Moderated)C++ Language (Moderated)C++ language (comp.lang.c++)C++ language, library and standards Non-English C/C++ Newsgroups----------------C++ (German)C++ (French)  You cannot post new topics in this forum You cannot reply to topics in this forum You cannot edit your posts in this forum You cannot delete your posts in this forum You cannot vote in polls in this forum Powered by phpBB © 2001, 2006 phpBB Group SEO toolkit © 2004-2006 webmedic.