mikpe@mina.liu.se (Mikael Pettersson) (11/28/88)
Consider the following program: ---snip--- #include <stdio.h> int count = 0; void bletch(int junk) { printf("count == %d\n", count); /* 0 or 1 ? */ } main() { bletch(count++); exit(0); } ---snip--- This program exemplifies the situation where the behaviour of the callee depends on whether the caller has completed *all* of the side effects from evaluating the arguments of the call or not. Q: what sayeth the dpANS about this? If count++ _is_defined_as_ (++count, count-1) then I guess things are cool, but what if the definition is less strict? /Mike -- Mikael Pettersson ! Internet:mpe@ida.liu.se Dept of Comp & Info Science ! UUCP: mpe@liuida.uucp -or- University of Linkoping ! {mcvax,munnari,uunet}!enea!liuida!mpe Sweden ! ARPA: mpe%ida.liu.se@uunet.uu.net
chris@mimsy.UUCP (Chris Torek) (11/30/88)
In article <1077@mina.liu.se> mikpe@mina.liu.se (Mikael Pettersson) writes: >Consider the following program: [edited] >void bletch(int junk) { printf("count == %d\n", count); /* 0 or 1 ? */ } >main() { bletch(count++); >This program exemplifies the situation where the behaviour of the >callee depends on whether the caller has completed *all* of the >side effects from evaluating the arguments of the call or not. >Q: what sayeth the dpANS about this? The dpANS sayeth that the program is to print `1'. All side effects are to have completed by the next `sequence point'. The list of sequence points includes comma expressions, `&&', `||', and not least, function calls. Obviously, statement bounardies (semicolons) are sequence points as well. >If count++ _is_defined_as_ (++count, count-1) then I guess things >are cool, but what if the definition is less strict? The definition is less strict, but the side effect is done by the next sequence point. Insert a sequence point and your side effect is guaranteed; without it, it is not. -- In-Real-Life: Chris Torek, Univ of MD Comp Sci Dept (+1 301 454 7163) Domain: chris@mimsy.umd.edu Path: uunet!mimsy!chris
evil@arcturus.UUCP (Wade Guthrie) (12/03/88)
In article <14758@mimsy.UUCP>, chris@mimsy.UUCP (Chris Torek) writes: > . . . All side effects > are to have completed by the next `sequence point'. The list of > sequence points includes comma expressions, `&&', `||', and not > least, function calls. Obviously, statement bounardies (semicolons) > are sequence points as well. I believe that binary operators such as '-' are also `sequence points'. Doesn't this cause the behavior of: a = b++ - b++; to be undefined (different answers depending on right-to-left evaluation versus left-to-right)? Wade Guthrie Rockwell International Anaheim, CA (Rockwell doesn't necessarily believe / stand by what I'm saying; how could they when *I* don't even know what I'm talking about???)
chris@mimsy.UUCP (Chris Torek) (12/04/88)
In article <2916@arcturus> evil@arcturus.UUCP (Wade Guthrie) writes: >I believe that binary operators such as '-' are also [dpANS] `sequence points'. No, none of +, -, *, /, or % introduce a sequence point. >Doesn't this cause the behavior of: > > a = b++ - b++; > >to be undefined (different answers depending on right-to-left evaluation >versus left-to-right)? You must be confused about sequence points. The order of evaluation of objects on the left and right sides of the arithmetic operators listed above is undefined. The *only* purpose for a sequence point is to guarantee side effects, so a sequence point within an operation whose evaluation order is undefined would be, er, `pointless'. -- In-Real-Life: Chris Torek, Univ of MD Comp Sci Dept (+1 301 454 7163) Domain: chris@mimsy.umd.edu Path: uunet!mimsy!chris