Confused over piece of code in parse_room()

Castillo wrote: Correction the code is:

CREATE(new_descr, struct extra_descr_data, 1);
      new_descr->keyword = fread_string(fl, buf2);
      new_descr->description = fread_string(fl, buf2);

Some code
Some code
{ <=====
Some code
Some code
} <======
Some code
Some code

thomas wrote: Yes, that's a needed fix for NULL from fread_string().

DESCRIPTION
The strchr() function locates the first occurrence of the character c
(converted to a char) in the string s. The terminating NUL character is
considered part of the string. If c is ‘\0’, strchr() locates the
terminating ‘\0’.

The index() function is an old synonym for strchr().

RETURN VALUES
The strchr() function returns a pointer to the located character or NULL
if the character does not appear in the string.

thomas wrote: And, in case someone needs to search for it, { } is called a scope.

It does something - it limits the scope of any variables to this part of the code. In C, you can only define new variables at the start of a scope. So this scope is here to allow us to create the end variable.

thomas wrote: It also means that this is a bit hacky. Had I made code like it at work, my colleagues might have stopped me and instead insisted on making me create a function for this scope. I'll see if I can have a look at improving this later.

It does something - it limits the scope of any variables to this part of the code. In C, you can only define new variables at the start of a scope. So this scope is here to allow us to create the end variable.

In C, you can only define new variables at the start of a scope.

case 'E':
      CREATE(new_descr, struct extra_descr_data, 1);
      new_descr->keyword = fread_string(fl, buf2);
      new_descr->description = fread_string(fl, buf2);
      /* Fix for crashes in the editor when formatting. E-descs are assumed to
       * end with a \r\n. -Welcor */
  //    {
        char *end = strchr(new_descr->description, '\0');
        if (end > new_descr->description && *(end-1) != '\n') {
          CREATE(end, char, strlen(new_descr->description)+3);
          sprintf(end, "%s\r\n", new_descr->description); /* snprintf ok : size checked above*/
          free(new_descr->description);
          new_descr->description = end;
        }
   //   }
      new_descr->next = world[room_nr].ex_description;
      world[room_nr].ex_description = new_descr;
      break;

*end is mallocated that is 3 bits bigger than the original

So this tells me that if the preceding code in that case statement is executed then the contents between the braces must be executed.

Which is why new_descr->description must be freed and reassigned to *end as the life of *end as a variable only exists within the scope defined by that set of braces, but new_descr->description is a variable exits in an external scope?

Castillo wrote: 3 bytes :)
(bits are for binary 01010010

thomas wrote: No, what is actually happening is that we're reusing the end variable.

First off - you need to understand that a string in C is "just a byte array". There's nothing fancy about it [3]. It contains chars by their ASCII byte value. To signify when the string ends, we put in a 0 byte. So the string "hi" will be represented by the char array [104, 105, 0]. The character notation in C uses single quotes and backslashes to signify the same thing: "hi" = 'h','i','\0' = '\104','\105','\0'
So all non-NULL strings contain at least one character - the 0 byte. Using the same annotation, the empty string ("") looks like this : '\0'.

        }
      } // break out of scope. The "end" variable no longer exists.

Castillo wrote: The bigger and new string created by CREATE(end, char, strlen(new_descr->description)+3);
is also permanent memory.
so, new_descr->description will point to a permanent memory string.
It's doesn't matter after that if the variable 'end' is destroyed.
new_descr->description knows that address of the string memory. And, that isn't destroy, neither the permanent memory.

thomas wrote: [1] Reusing is also hacky. As you a evidence of,