kpk@gitpyr.UUCP (09/17/86)
Who among us has not been bitten at least once by writing if (x = 1) ... when he/she meant if (x == 1) What I want to know is Has anyone written a program to scan a C program and issue a warning message for those lines where an assignment has occured in a conditional statement (i.e. if, while, the third statement in a for loop)? If anyone has written such a program, why not post it (or at least mail me a copy)? To any yacc and lex hacks: why not see if you can whip up such a program?
rbutterworth@watmath.UUCP (Ray Butterworth) (09/18/86)
> Who among us has not been bitten at least once by writing > if (x = 1) ... > when he/she meant > if (x == 1) > What I want to know is > > Has anyone written a program to scan a C program and issue a warning message > for those lines where an assignment has occured in a conditional statement > (i.e. if, while, the third statement in a for loop)? There is already such a program: lint. Here's the modified section you want from the BSD 4.2 version of /usr/src/usr.bin/lint/lpass1.c. contx( p, down, pl, pr ) register NODE *p; register *pl, *pr; { *pl = *pr = VAL; if (p->in.type==UNDEF) down=VAL; /* (void) cast */ switch( p->in.op ){ case NOT: *pl=down; case ANDAND: case OROR: if (hflag&&(p->in.right->in.op==ASSIGN)) werror("Possible unintended assignment"); case QUEST: *pr = down; case CBRANCH: if (hflag&&(p->in.left->in.op==ASSIGN)) werror("possible unintended assignment"); break; case SCONV: case PCONV: ...
cg@myrias.UUCP (Chris Gray) (09/18/86)
The proper place for this check is of course as a warning issued by the C compiler. It's not all that hard - it took me about 30 lines of code in the compiler (ANSI draft based) that I've done. It's a check I felt was quite important, since I've been bitten by it too many times (I program in other languages besides C, where '=' IS a comparison). Chris Gray (...alberta!myrias!cg)
caesar@ge-dab.UUCP (Robert J. Caesar Jr.) (09/21/86)
I don't have a programs, but this might help as a future solution. In the project I am involved with, we use: #define is == #define isn_t != This has helped us avoid a lot of headaches. The code reads a little easier too. if (x is 1) if (y isn_t x) -- Bob Caesar General Electric SCSD caesar@static.dab.ge.com PO Box 2500 Room 4336 (904) 258-3043 Daytona Beach, Florida 32015 ...!mcnc!ge-rtp!ge-dab!caesar
Eliel@smoke.UUCP (09/24/86)
This is a good question, although I think that for CMS/SP users, using yacc
or lex may be a bit overboard when XEDIT provides the "all" command that
would let one easily scan for that kind of error. I suppose that a macro
to do this before compiling would be easy to write. As I don't have time
right now, let me outline what I mean:
in CW EXEC :
"XEDIT "fn ft/* h or c */" (PROFILE chk_equ"
if rc ^= 0 then exit /* flag indicating desire to continue or no*/
else CW ....
chk_equ XEDIT would have something like this :
"all /for/ && /=/ "
some search code...
"all /while/ && /=/ "
some more..., etc, etc.
Let me know if any one has the time to do it so that I can get a copy!
also, if any one knows of versions of yacc and/or lex for the CMS/SP environ-
ment, please let me know as well!
have a day,
elielbrett@wjvax.UUCP (Brett Galloway) (09/25/86)
In article <270@myrias.UUCP> cg@myrias.UUCP (Chris Gray) writes: >The proper place for this check is of course as a warning issued by the C >compiler. It's not all that hard - it took me about 30 lines of code in the >compiler (ANSI draft based) that I've done. It's a check I felt was quite >important, since I've been bitten by it too many times (I program in other >languages besides C, where '=' IS a comparison). I disagree; the proper place for this check is NOT as a warning issued by the compiler. "if(var=value)" is a perfectly valid, supported, and documented construction, endorsed by K & R et al. The fact that it is misleading is an artifact of the ambiguity of the meaning of "equal" and "=". In any event, it is a well-defined and supported feature of the language. It is ridiculous for the compiler to bark on the expression (var=value) when it happens to fall into the context of a simple if(). The proper place for this check is within lint, whose purpose is to ... detect features of ... [C programs] ... which are likely to be bugs, or non-portable, or wasteful. - lint(1), from 4.2 BSD manual Further, it should be a special test by lint, invoked by a special argument, so that those of us who like to use lint don't have to listen to the warning forever. -- ------------- Brett Galloway {pesnta,twg,ios,qubix,turtlevax,tymix,vecpyr,certes,isi}!wjvax!brett