C++Talk.NET C++ language newsgroups   Archives   FAQ   Search   Memberlist   Usergroups   Register   Profile   Log in to check your private messages   Log in  Evaluation precedence questions Goto page 1, 2, 3, 4  Next          C++Talk.NET Forum Index -> C++ Language (Moderated) View previous topic :: View next topic   Author Message Alessandro Molina Guest Posted: Wed Jan 14, 2004 11:08 pm    Post subject: Evaluation precedence questions Standard asserts that there isn't an official way to evaluate statements like: a[i] = i++ and similars I was just curious about the reasons. Why didn't the standard define a generic evaluation direction for expressions? What problems will happen stating that compilers must always evaluate expressions from left or from right for example? [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Hyman Rosen Guest Posted: Thu Jan 15, 2004 10:30 am    Post subject: Re: Evaluation precedence questions Alessandro Molina wrote: Quote: I was just curious about the reasons. Why didn't the standard define a generic evaluation direction for expressions? Because We've Always Done It This Way (tm) People will blather on about optimization possibilities, but that's a crock. It would be completely reasonable to adopt Java's order of evaluation semantics and finally get rid of this stupid albatross. But it will never happen. [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Jack Klein Guest Posted: Thu Jan 15, 2004 10:38 am    Post subject: Re: Evaluation precedence questions On 14 Jan 2004 18:08:07 -0500, Alessandro Molina <javanx (AT) supereva (DOT) it> wrote in comp.lang.c++.moderated: Quote: Standard asserts that there isn't an official way to evaluate statements like: a[i] = i++ and similars I was just curious about the reasons. Why didn't the standard define a generic evaluation direction for expressions? For compatibility with C, and for performance. Quote: What problems will happen stating that compilers must always evaluate expressions from left or from right for example? What will happen is this: Any order that you pick will generate efficient object code on some compilers for some processors, and inefficient object code on some other compilers for some other processors. One of the basic tenets of C++ is to be as efficient as C for low-level programming, and in some cases more efficient than C for high-level programming. One of the basic tenets of C, and so inherited by C++, is to be as efficient as possible. Period. There are languages that will provide the guarantees that you want. In general, they are not as fast or efficient as C or C++ at the kind of low-level programming that C and C++ are used for, especially in systems programming and real-time embedded systems. -- Jack Klein Home: http://JK-Technology.Com FAQs for comp.lang.c http://www.eskimo.com/~scs/C-faq/top.html comp.lang.c++ http://www.parashift.com/c++-faq-lite/ alt.comp.lang.learn.c-c++ http://www.contrib.andrew.cmu.edu/~ajo/docs/FAQ-acllc.html [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Dhruv Guest Posted: Thu Jan 15, 2004 10:40 am    Post subject: Re: Evaluation precedence questions On Wed, 14 Jan 2004 18:08:07 -0500, Alessandro Molina wrote: Quote: Standard asserts that there isn't an official way to evaluate statements like: a[i] = i++ and similars I was just curious about the reasons. Why didn't the standard define a generic evaluation direction for expressions? What problems will happen stating that compilers must always evaluate expressions from left or from right for example? I don't know much about the issues, but I think that the example you have given above should be legal. Consider: i = 0; Then, the quiivalen code generated would be: (possibly!) a[0] = 0; ++i; Refards, -Dhruv. [ 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: Thu Jan 15, 2004 3:12 pm    Post subject: Re: Evaluation precedence questions In message <gg3c00p4aqehbmvo3bvjd94ande68cnqd9 (AT) 4ax (DOT) com>, Jack Klein <jackklein (AT) spamcop (DOT) net> writes Quote: On 14 Jan 2004 18:08:07 -0500, Alessandro Molina wrote in comp.lang.c++.moderated: Standard asserts that there isn't an official way to evaluate statements like: a[i] = i++ and similars I was just curious about the reasons. Why didn't the standard define a generic evaluation direction for expressions? For compatibility with C, and for performance. I recently suggest at a WG14 meeting (C Standard) that future releases of the C Standard should specify the order of evaluation (it cannot break any existing well-formed code because the order is currently unspecified). The only problem is (as I knew when I raised the question) that it would break some existing compilers. -- Francis Glassborow ACCU Author of 'You Can Do It!' see http://www.spellen.org/youcandoit [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Hyman Rosen Guest Posted: Fri Jan 16, 2004 2:53 am    Post subject: Re: Evaluation precedence questions Jack Klein wrote: Quote: For compatibility with C Of course not, since giving defined behavior to formerly undefined code cannot cause incompatibility. and for performance. In older threads on the same subject people have managed to come up with isolated cases where God can generate better code by using an order of evaluation other than "left before right, operands before operation". The practical effects of these, especially compared with the decades of erroneous code that this undefined behavior has engendered, are nil. [ 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: Fri Jan 16, 2004 3:01 am    Post subject: Re: Evaluation precedence questions On 15 Jan 2004 05:40:21 -0500, "Dhruv" <dhruvbird (AT) gmx (DOT) net> wrote: Quote: On Wed, 14 Jan 2004 18:08:07 -0500, Alessandro Molina wrote: a[i] = i++ and similars I don't know much about the issues, but I think that the example you have given above should be legal. Consider: It is legal, well formed. You just have no idea what it will do. Quote: i = 0; Then, the quiivalen code generated would be: (possibly!) a[0] = 0; ++i; or ++i; a[1] = 0; or I is a 32 bit thing on an 8 bit machine and the software increment is interleaved with the subscripting and formats the hard drive. John [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Joshua Lehrer Guest Posted: Fri Jan 16, 2004 1:07 pm    Post subject: Re: Evaluation precedence questions Jack Klein <jackklein (AT) spamcop (DOT) net> wrote Quote: Any order that you pick will generate efficient object code on some compilers for some processors, and inefficient object code on some other compilers for some other processors. so rather than pick one, which will be efficient on some platforms, and not on others, we pick neither, which means that you can't write the code at all? joshua lehrer factset research systems NYSE:FDS [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Frank Birbacher Guest Posted: Fri Jan 16, 2004 1:08 pm    Post subject: Re: Evaluation precedence questions Hi! Dhruv wrote: Quote: I don't know much about the issues, but I think that the example you have given above should be legal. Consider: It is not legal if i is a built in type (int, long, ...). Frank [ 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: Sat Jan 17, 2004 1:25 am    Post subject: Re: Evaluation precedence questions In message <ngdd00p1nmcbhg3kb3p7bvb8grvksjgfkq (AT) 4ax (DOT) com>, John Potter <jpotter (AT) falcon (DOT) lhup.edu> writes Quote: On 15 Jan 2004 05:40:21 -0500, "Dhruv" <dhruvbird (AT) gmx (DOT) net> wrote: On Wed, 14 Jan 2004 18:08:07 -0500, Alessandro Molina wrote: a[i] = i++ and similars I don't know much about the issues, but I think that the example you have given above should be legal. Consider: It rather depends on what is meant by 'legal'. It breaches a requirement of the Standard (no diagnostic required) in that it evaluates i for two distinct purposes only one of which is to determine the value to be written (to i) -- Francis Glassborow ACCU Author of 'You Can Do It!' see http://www.spellen.org/youcandoit [ 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: Sat Jan 17, 2004 1:26 am    Post subject: Re: Evaluation precedence questions In message <1074174627.75607 (AT) master (DOT) nyc.kbcfp.com>, Hyman Rosen <hyrosen (AT) mail (DOT) com> writes Quote: Jack Klein wrote: For compatibility with C Of course not, since giving defined behavior to formerly undefined code cannot cause incompatibility. Careful about the terminology, undefined is usually applied to code where the Standard says nothing at all and so it can do anything the compiler decides. Order of evaluation is 'unspecified' which means that the compiler is strongly constrained and cannot do anything, only choose some order of evaluation that will deliver the values of operands in a timely fashion. -- Francis Glassborow ACCU Author of 'You Can Do It!' see http://www.spellen.org/youcandoit [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Ben Hutchings Guest Posted: Sat Jan 17, 2004 1:32 am    Post subject: Re: Evaluation precedence questions Hyman Rosen wrote: Quote: Jack Klein wrote: For compatibility with C Of course not, since giving defined behavior to formerly undefined code cannot cause incompatibility. and for performance. In older threads on the same subject people have managed to come up with isolated cases where God can generate better code by using an order of evaluation other than "left before right, operands before operation". snip Nonsense. If the implementation is free to evaluate operands in any order it can use this freedom to minimise the number of registers or stack slots used and thus to reduce memory traffic. This is a fairly basic optimisation that was covered in the Compilers and Operating Systems course I took as an undergraduate. [ 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: Sat Jan 17, 2004 1:33 am    Post subject: Re: Evaluation precedence questions In message <31c49f0d.0401151233.3c6de912 (AT) posting (DOT) google.com>, Joshua Lehrer <usenet_cpp (AT) lehrerfamily (DOT) com> writes Quote: Jack Klein <jackklein (AT) spamcop (DOT) net> wrote Any order that you pick will generate efficient object code on some compilers for some processors, and inefficient object code on some other compilers for some other processors. so rather than pick one, which will be efficient on some platforms, and not on others, we pick neither, which means that you can't write the code at all? How does that follow? By making some things unspecified we turn them into QoI issues. -- Francis Glassborow ACCU Author of 'You Can Do It!' see http://www.spellen.org/youcandoit [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Hyman Rosen Guest Posted: Sat Jan 17, 2004 3:13 am    Post subject: Re: Evaluation precedence questions Frank Birbacher wrote: Quote: It is not legal if i is a built in type (int, long, ...). And so we spin this stupid thread again and again and again... Am I the only one who is sick to death of it? [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Dhruv Guest Posted: Sat Jan 17, 2004 3:20 am    Post subject: Re: Evaluation precedence questions On Thu, 15 Jan 2004 22:01:03 -0500, John Potter wrote: Quote: On 15 Jan 2004 05:40:21 -0500, "Dhruv" <dhruvbird (AT) gmx (DOT) net> wrote: On Wed, 14 Jan 2004 18:08:07 -0500, Alessandro Molina wrote: a[i] = i++ and similars I don't know much about the issues, but I think that the example you have given above should be legal. Consider: It is legal, well formed. You just have no idea what it will do. [...] Quote: ++i; a[1] = 0; or I is a 32 bit thing on an 8 bit machine and the software increment is interleaved with the subscripting and formats the hard drive. Ok, so let me get a few things straight in my mind: 1. Is = a sequence point? 2. wrt. operator++(int) [post increment] applied to ints (PODS), when exactly is it said that the increment will happen? Possible options include: A) Between the assignment taking place, and the RHS expression being evaluated. B) After the value of i has been obtained in the expression, so something like this would be illegal foo (i, i++) (Which I guess is illegal currently). C) After the complete expression has been evaluated. (Which was my initial assumption). D) None of the above? Regards, -Dhruv. [ 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 Goto page 1, 2, 3, 4  Next Page 1 of 4   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.