Please help with this C syntax error

[dietert1]

Some embedded development environments support inclusion of data files at link time. I remember using that method with IAR EWArm to include HTML files, javascripts, images and style sheets of a web server. One gives to the linker the file path, a label to reference the base address in the C sources and a section id.

Regards, Dieter

[peter-h]

Here is another mystery syntax error



SGBOOL is uint8_t
SGUINT is uint16_t
GCONSTP is const

GPALETTE_RGB is
typedef struct _GPALETTE_RGB
   {
   SGUCHAR r;
   SGUCHAR g;
   SGUCHAR b;
   } GPALETTE_RGB;

PFCODE is
  /* Memory type qualifier used for pointer to fixed data (if GCODE var * is not enough) */
  #define PFCODE   /* nothing */

[radiolistener]

try to use typedef instead of preprocessor #defines

[peter-h]

The #defines for e.g. uint8_t are everywhere in this bit of code (not mine). The problem is elsewhere and may be in the typedef'd struct.

[paulca]

In Uni when it came to C and C++ we were forbidden to do many things.

One of them was to use contractions.  The reason being,

for( ; AMACRO++; AMACRO<limit )

Becomes a completely random bit of code once the preprocessing has finished with it.  It might, if you are lucky produce a compile error.  If you are unlucky it will do something that appears valid, most of the time.

Using
for( ; AMACRO=AMACRO+1; AMACRO<limit )

Will catch more cases of inappropriate use of AMACRO than the former will.  You are first ensuring the AMACRO can "store" a result in the first place.  If it's a function call wrapper, then ++ will either post increment a register that will immediately be discarded or it will bomb out with a wicked compiler error.


MACRO_DEFINED_FUNCTION( parameters ) -1U

Is this asking for trouble? and relying exclusively on how that MACRO is implemented and if it was ever intended to be used that way.

In C++ it gets worse when operators can be macro'd!

The way to debug these things is unfortunately to add the "keep preprocessing output" flag and go and look at what ACTUALLY got compiled.

I don't know if it solves your problem, but just a "word from the wise", (my uni, not me).

[peter-h]

I am also surprised you can have a #define of nothing.

[dietert1]

This may be a missing bracket above within the same source file. I mean the function doesn't have a certain syntax error but a function wasn't expected..

[ejeffrey]

I am also surprised you can have a #define of nothing.

You probably aren't if you think about it.  It's extremely common.  It's mostly used along side #ifdef, so the macros are usually never expanded.   For instance the standard pattern of #include guards doesn't specify a value.

It's also common when you need something to conditionally expand.  This is frequently used when you need non-standard qualifiers on some platforms.  For instance windows DLLs require functions to have the dllexport storage class attribute so the following pattern is common:

Code: [Select]

#ifdef WINDOWS
#define DLLEXPORT dllexport
#else
#define DLLEXPORT
#endif

DLLEXPORT int foo(int x);

You can use the same pattern for different calling conventions (cdecl vs fastcall vs isr), section attributes, or anything else that would be an error in some situations.

[gamalot]

I am also surprised you can have a #define of nothing.

It is very common in include garding.

Code: [Select]

#ifndef FOO_H
#define FOO_H

......

#endif /* FOO_H */