I have having following warning in gcc compilation in 32 bit architecture but not having any such warning in 64 bit architecture

symbol.c: In function 'symbol_FPrint':

symbol.c:1209: warning: format '%ld' expects type 'long int', but argument 3 has type 'SYMBOL'
symbol.c: In function 'symbol_FPrintOtter':
symbol.c:1236: warning: format '%ld' expects type 'long int', but argument 3 has type 'SYMBOL'
symbol.c:1239: warning: format '%ld' expects type 'long int', but argument 3 has type 'SYMBOL'
symbol.c:1243: warning: format '%ld' expects type 'long int', but argument 3 has type 'SYMBOL'
symbol.c:1266: warning: format '%ld' expects type 'long int', but argument 3 has type 'SYMBOL' 

In symbol.c

1198 #ifdef CHECK
1199     else {
1200       misc_StartErrorReport();
1201       misc_ErrorReport("\n In symbol_FPrint: Cannot print symbol.\n");
1202       misc_FinishErrorReport();
1203     }
1204 #endif
1205   }
1206   else if (symbol_SignatureExists())
1207     fputs(symbol_Name(Symbol), File);
1208   else
1209     fprintf(File, "%ld", Symbol);
1210 }

And SYMBOL is defined as:

typedef size_t SYMBOL

When i replaced '%ld' with '%zu' , i got the following warning:

symbol.c: In function 'symbol_FPrint':
symbol.c:1209: warning: ISO C90 does not support the 'z' printf length modifier

Note: From here it has been edited on 26th of march 2010 and and following problem has beeen added because of its similarity to the above mentioned problem.

I have following statement:

printf("\n\t %4d:%4d:%4d:%4d:%4d:%s:%d", Index, S->info, S->weight,
       Precedence[Index],S->props,S->name, S->length);

The warning I get while compiling in 64 bit architecture is :

format ‘%4d’ expects type ‘int’, but argument 5 has type ‘size_t’

here are the definitions of parameter:

  NAT    props;
  typedef  unsigned int     NAT;

How can i get rid of this so that i can compile without warning in 32 and 64 bit architecture?

What can be its solution?


Use %zu rather than %ld as the format for size_t and then you'll get correct behaviour (and no warnings) in both 32-bit and 64-bit builds.

If for some reason you can't use %zu (e.g. old or non-standard compiler), then you can do something like this:

#ifdef __LP64__ // if 64 bit environment
#define FMT_SIZE_T "llu"
#define FMT_SIZE_T "lu"

Then when you need to use printf with somethign of type size_t you can do this:

printf("(sizeof(void *) = %"FMT_SIZE_T" bytes \n", sizeof(void *));

i find %lu to be the only C89-compliant (or, gcc-c89-warnings-compliant) format for size_t, on linux. under glibc, size_t seems to always be long unsigned int, so %lu is appropriate there.

in particular i found that gcc-4.4.3 with -std=c89 -pedantic or -std=c89 -Wall will warn about %llu and %lld. makes sense, since they expect long long, a C99 type.

tragically there isn't a way to print a size_t in a platform-independent C89-compliant way. in C89, all printf formats refer to specific primitive types (long, int, etc), but in all C dialects, size_t is platform-dependent (might be long or int or something else).

as usual, microsoft throws you its own special monkey wrench. the windows compilation environment is a badly-modified C89: it has long long, but no ll printf modifier. in windows-land, %llu becomes %I64u and %zu becomes %Iu.

i ended up using the following in my lab's codebase:

#ifdef _WIN32
#define PRIuZ "Iu"
#define PRIuZ "lu"


printf("%"PRIuZ"\n", sizeof(struct foo));

i modeled the name from the definitions in C99's inttypes.h.

aside: my version of win64 mingw-gcc (4.4.5 20100527 (prerelease) [svn/rev.159909 - mingw-w64/oz]), plus -Wall or -pedantic, always warns when printing a size_t. it complains that %I64u and %Iu are nonstandard, and that %lu is incorrect. seems like the compiler is correctly getting tripped up by microsoft's insanity. i had to use -pedantic -Wformat=0 to get it to not whine about %Iu.

Need Your Help

NSBundle pathForResource failing in shell tool

objective-c cocoa macos

I've noticed some weird behavior with NSBundle when using it in a

nested model form and update_attributes

ruby-on-rails attributes update-attributes

I have a nested-model form with a one-to-many relationship between a class Project and class TeamMember, and in the controller, an update function like :