Chris Green chris@areti.co.uk writes:

Thanks, yes, I've done similar tricks with #define when wanting a set of enum constants to match a set of string variables. The above certainly does address the globals issue. However it does mean that you have values (other than constants) in the .h file which is (sort of) a bit naughty.

I don't see why...?

On Thu, Aug 11, 2005 at 12:51:49PM +0100, Chris Green wrote:

You could of course make all the actual definitions one huge #define and use it in two places with your two sets of EXTERN and INIT defines set differently. This is essentially what I did for my enum/string one.

... and going down this route I have come up with:-

#define CLOCK \ {\ VAR (int, xsize, 54)\ VAR (int, ysize, 54)\ VAR (char, bgcolor[100], "#123456")\ VAR (char, fgcolor[100], "#FFFFFF")\ VAR (bool, iconic, false)\ }

#define VAR(a, b, c) a b;

struct Clk CLOCK; // declare the structure

#undef VAR #define VAR(a, b, c) c,

struct Clk clk = CLOCK; // intialiase the structure

int main(int argc, char * argv[]) { int i; }

Using 'CC -E' (this is a Solaris box) shows the output from the pre-processor to be a correctly declared and initialised struct.

On Thu, Aug 11, 2005 at 11:27:27AM +0100, Stuart Bailey wrote:

...

Is there a good way around this problem or have I just got to bite the bullet and decide on one or other of the above and go with it?

I am really a C++ developer, but with some C experience.

Really, any coding style should make the code easier to read, and easier to maintain - this includes documentation.

I normally follow these rules for using globals and structures:

Only use globals if absolutely necessary (always try to pass data as arguments to functions, either as values or pointers)

Quite agree which is why I wasn't keen on the 'global' solution.

If a variable is required to remain in scope for multiple calls to a function, rather than make it global, make it static in the function that's using it.

Yes, if that's the intention then that's the way to do it.

Use structures to group related items together. So a structure containing say screen attributes would be OK, but just grouping variables together for the sake of it would not be acceptable.

Which is one of the reasons for not having a huge 'config' structure in my original question.

However none of this addresses my question really. The issue is trying to avoid repeating what is essentially the same thing over and over again when you have a large number of loosely associated values (the clock configuration parameters).

If they're globals you have to both initialise them and declare them as extern. Every global is (at least) duplicated as:-

int xsize = 1234; // in a .c file somewhere

and

extern int xsize; // in a .h file somewhere

If (as there should be) there are very few globals this is OK, but with the 50 or more the clock requires it's messy. They *are* required in different places because they're modified by the command line parsing and used in the X code which are separate source files.

If they were passed around as function parameters you'd end up with huge parameter lists to the functions, unless you move to the alternative struct .....

If you use a struct to contain the configuration you avoid many of the nasties of globals, however initialising the values is messy IMHO. You get stuff like:-

In a header file:-

struct ClockConfig { int xsize; int ysize; char bgcolor[COL_SIZE]; char fgcolor[COL_SIZE]; bool iconic; ... ... ... etc. };

and then, in the code somewhere:-

struct ClockConfig clockCfg = { 54, 54, "#123456", "#FFFFFF", FALSE, ... ... .. etc. };

Which, to my mind, is truly horrible and difficult to maintain because you have to be very careful to maintain the one-to-one correspondence between the initialisation values and the struct. Commenting will of course help but it again represents duplicated information which really shouldn't be necessary.

Stuart Bailey stuart@linusoft.co.uk writes:

Yes, but as far as I can remember (and I may be verging on C++), you can have functions within a struct. Therefore, you could create an initialisation function, with well named parameters (with defaults), that will setup the struct for you. Although you will still have a long parameter list. Unfortunately, I can't remeber the syntax - and I can't find any of my C books to reference. But basically, a C++ class is essentially a structure whose members are by default private rather than public as in structures.

You are thinking of a constructor, which is part of C++, not of C. You could indeed use that to tie names to initial values, though it still has the problem that you have to list the names twice, opening up scope for error.

GCC/C99 designated initializer syntax avoids the C++ dependency but has the same flaw.

---

This email and any files transmitted with it are confidential. If you are not the intended recipient, please email postmaster@linusoft.co.uk immediately. You should not copy or use this email or attachments for any purpose nor disclose their contents to any other person.

NO BINDING CONTRACT WILL RESULT FROM THIS E-MAIL UNTIL SUCH TIME AS A WRITTEN DOCUMENT IS SIGNED ON BEHALF OF LinuSoft.

LinuSoft cannot accept any responsibility for the completeness or accuracy of this message as it has been transmitted over public networks.

---

This is very silly.

Coding styles can be an emotive subject (along the lines of which is better, vi or emacs). I for one find the GNU recommended style to be horrible, preferring the guidance found in Documentation/CodingStyle (from any kernel source) to be reasonable.

On Thursday 11 August 2005 13:25, Stuart Bailey wrote:

Yes, but as far as I can remember (and I may be verging on C++), you can have functions within a struct. Therefore, you could create an initialisation function, with well named parameters (with defaults), that will setup the struct for you. Although you will still have a long parameter list. Unfortunately, I can't remeber the syntax - and I can't find any of my C books to reference. But basically, a C++ class is essentially a structure whose members are by default private rather than public as in structures.

struct Foo{ public: Foo() {a=0; b=2; c=3; bar="idunno"};

int a; int b; int c; char bar[80]; }

However, this only works with C++, so one small trick I use is a macro defined in the same header as the struct definition..

typedef struct { double s, x, y, z; } PmRotation;

#define PmRotation_init {1, 0, 0, 0}

In the source, it is a simple matter of doing:

PmRotation foo = PmRotation_init;

Once again, this gets messy if the multitude of structs needed different initialisation values...

Which is better - That really depends on how far you want to go in embracing C++ and OOP. Certainly C++ allows you to hide and protect many variables that would otherwise become global.

Regards, Paul.

On 11/08/05 13:07:10, Chris Green wrote:

However none of this addresses my question really. The issue is

trying

to avoid repeating what is essentially the same thing over and over again when you have a large number of loosely associated values (the clock configuration parameters).

If they're globals you have to both initialise them and declare them as extern. Every global is (at least) duplicated as:-

int xsize = 1234;   // in a .c file somewhere

and

extern int xsize;   // in a .h file somewhere

If (as there should be) there are very few globals this is OK, but with the 50 or more the clock requires it's messy. They *are* required in different places because they're modified by the command line parsing and used in the X code which are separate source files.

If you think globals are the best solution then you could adopt a code generation approach.

Create a file called globals.c or something similar and list the variables and their initialisers there, so in your above example:

int xsize = 1234;

Now use a script, perhaps in perl or similar, to process this file by taking each non-comment line, grabbing the part before the '=' sign (or before the final semicolon if there is no '=' on that line), prepend 'extern ' to it and write it out. The output of this script goes into globals.h or similar which can then be included by other modules.

This script can be run from a makefile to to add a new variable you edit globals.c and run make.

Steve.