4

I have the following code, which is not working as expected. It compiles, but throws a lot of warnings and segfaults when executed:

#include <stdio.h>

enum output {
    A,
    B,
    C,
    D,
};

struct translation {
    char *from;
    enum output to;
};

struct dictionary {
    struct translation *foo;
    struct translation *bar;
};

enum language {
    ONE,
    ANOTHER,
};

struct dictionary languages[] = {
        [ONE] = {
                .foo = {
                        {"LF", A},
                        {"LLF", C},
                        {"RRF", D},
                },
                .bar = {
                        {"L", B},
                },
        },
        [ANOTHER] = {
                .foo = {
                        {"FF", B},
                        {"RRF", D},
                },
                .bar = {
                        {"LF", B},
                        {"R", C},
                        {"RR", D},
                },
        },
};

int main(void)
{
        printf("%s\n", languages[ONE].foo[0].from);
        return 0;
}

I am probably initializing languages the wrong way.

  • I would like to have that languages array in which I can access different dictionaries by language: languages[ONE]
  • I would like to access then different translation tables with the dictionary field: languages[ONE].foo
  • All translation tables accessed with a language+field pair may have different array lengths, as shown in the code example

Is that even possible? What am I doing wrong?

When compiling with gcc I get this (cropped) output:

asdf.c:27:17: warning: braces around scalar initializer
                 .foo = {
                 ^
asdf.c:27:17: note: (near initialization for ‘languages[0].foo’)
asdf.c:28:25: warning: braces around scalar initializer
                         {"LF", A},
                         ^
asdf.c:28:25: note: (near initialization for ‘languages[0].foo’)
asdf.c:28:26: warning: initialization of ‘struct translation *’ from incompatible pointer type ‘char *’ [-Wincompatible-pointer-types]

[...]

The same warnings/notes repeat for multiple parts of the code.

1
  • @FiddlingBits Actually, that could be a workaround! :-) I tried with [] (no size) but that was illegal. You can post that as an answer if you want. Although I will wait before accepting to see if there is a way of making this work without having to specify a dictionary table size "big enough" to hold all translations. Commented Nov 19, 2018 at 21:35

2 Answers 2

3

Here are two things you can do:

  1. Allocate memory for struct translation *foo; and struct translation *bar; (you can also use malloc to dynamically allocate memory). For example:
struct dictionary
{
    struct translation foo[10];
    struct translation bar[10];
};
  1. Use a compound literal in your definition:
struct dictionary languages[] = {
    [ONE] = {
            .foo = (struct translation []){
                    {"LF", A},
                    {"LLF", C},
                    {"RRF", D},
            },
            .bar = (struct translation []){
                    {"L", B},
            },
    },
    [ANOTHER] = {
            .foo = (struct translation []){
                    {"FF", B},
                    {"RRF", D},
            },
            .bar = (struct translation []){
                    {"LF", B},
                    {"R", C},
                    {"RR", D},
            },
    },
};

Note

As mentioned by @M.M, adding the qualifier const before struct dictionary is a good idea if its values won't change during runtime.

Sign up to request clarification or add additional context in comments.

9 Comments

Love the latter <3. I am learning a lot of C with this project. :-D
something like struct translation foo[10]; (leaving some of them uninitialized) is a very poor idea while compound literals are feature of C99 language
@VTT Sure. I mentioned it as an option, not a recommendation.
@VTT why is it a poor idea in your opinion?
@VTT not necessarily; in some systems memory is cheap and other concerns such as code clarity, portability, ease of development, ease of runtime access etc. may have a higher priority
|
1

Just initialize each array separately:

#include <stdio.h>

enum output
{
    a
,   b
,   c
,   d
};

struct translation
{
    char const * from;
    enum output  to;
};

struct dictionary
{
    struct translation * foo;
    struct translation * bar;
};

enum language
{
    one
,   another
,   languages_count
};

struct translation one_language_foo_translations[] =
{
    {"LF" , a}
,   {"LLF", c}
,   {"RRF", d}
};

struct translation one_language_bar_translations[] =
{
    {"L", b}
};

struct translation another_language_foo_translations[] =
{
    {"FF" , b}
,   {"RRF", d}
};

struct translation another_language_bar_translations[] =
{
    {"LF", b}
,   {"R" , c}
,   {"RR", d}
};

struct dictionary languages[languages_count] =
{
    {one_language_foo_translations, one_language_bar_translations}
,   {another_language_foo_translations, another_language_bar_translations}
};

int main(void)
{
    printf("%s\n", languages[one].foo[0].from);
    return 0;
}

online compiler

Comments

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.