[pcpro17]

Hello. I'm having trouble saving data to an ASCII file and then reading it back into the program. I'm able to save the data alright (I can view it in a text editor). However, I keep getting a segmentation fault when I try to read it back into the program.

Here's how I save the data:

   file_ptr = fopen( file_name, "w" );
      fprintf( file_ptr, "avg_get_time\t%u\0", values.avg_get_time );
      fprintf( file_ptr, "avg_loop_time\t%llu\0", values.avg_loop_time );
      ret = fclose( file_ptr );

When I read the data back in (after the program is restarted) I use the following magic:

   file_ptr = fopen( file_name, "r" );

      /* retrieve data from file            */
      n = 0;
      printf( "Trying to copy data from file . . .\n" );
      while( (file_data[n] = fgetc( file_ptr )) != EOF ) {
         printf( "%c", file_data[n] );
         n++;
      }
      ret = fclose( file_ptr );

      /* parse saved data from file         */
      m = 0;
      printf( "\nTrying to parse data from file . . .\n" );
      /* THE CODE CRASHES HERE */
      for( n = strchr( file_data, '\t' )+1; file_data[n-1] != '\0'; n++ ) {
          ascii_num[m] = file_data[n];
          m++;
      }
      printf( "%s\n", ascii_num );

I keep getting a segmentation fault in the last for loop. I've tried using '\n' and '\0'. Does anyone have any other suggestions?

Thanks.

 

[Chicon]

[SIZE=2]
      for( n = strchr( file_data, '\t' )+1; file_data[[COLOR=Red]n-1[/COLOR]] != '\0'; n++ ) {
          ascii_num[m] = file_data[[COLOR=Red]n[/COLOR]];
          m++;
      }
[/SIZE]

In the last loop, you're passing the value \0 to the function ascii_num!
In the for statement, you've just to replace the n-1 expression by n.

 

[AGCurry]

There are a number of problems with your code, but the most fatal aspect of it is that strchr() does not return an int - it returns char *. You are using that value as an array index, which C will happily let you do, but it is a sure path to dumping core.

I'm not sure what chicon was trying to say, as ascii_num looks to be an array, not a function.

 

[Chicon]

AGCurry is right, the strchr() returns a char * and ascii_numb is not a function as I believed. Sorry, I only focused on the way the loop increment is handled.

 

[pcpro17]

Ok, I decided to handle this a little differently using fgets() and strtok(). This is what I have so far:

      fgets( file_data, s, file_ptr );
      strtok( file_data, "\t" );
      ascii_num = strtok( NULL, "\n" );

The man page for strtok indicates that NULL should be passed for the first arguement on subsequest calls, but I keep getting an "incompatible types in assignment" error on the last line there when I try to compile. The man page isn't much help here because it says that strtok returns a character pointer, and that is what ascii_num[16] is.

Any thoughts? Thanks.

 

[Chicon]

There is a good example of the strtok function on this page. The function returns a char * pointer too.

 

[AGCurry]

fgets( file_data, s, file_ptr );
strtok( file_data, "\t" );
ascii_num = strtok( NULL, "\n" );

If ascii_num is declared as char[16], you can use it as a char * on the right side of assignments but not on the left side. Try declaring it as "char *ascii_num" .

 

[pcpro17]

fgets( file_data, s, file_ptr );
strtok( file_data, "\t" );
ascii_num = strtok( NULL, "\n" );

If ascii_num is declared as char[16], you can use it as a char * on the right side of assignments but not on the left side. Try declaring it as "char *ascii_num" .

Um, I don't think that's quite right because that's how I'm using it in my code, and I've been able to verify that it's reading back the number properly from the file.

The problem now is the 64-bit number is not being stored properly to the file. (I'm not sure why it quit working). Here is the code I'm using to store the values:

      fprintf( file_ptr, "%u\t%ull", values.avg_get_time, values.avg_loop_time );

For example, if avg_loop_time is 23540235014, the value stored in the file is "2065398534ll". For some reason it seems that %ull is not recognised by fprintf(). Any suggestions?

Thanks.

 

[Chicon]

Just try this :
fprintf( file_ptr, "%l\t%ll", values.avg_get_time, values.avg_loop_time );

refered documentation

 

[pcpro17]

Just try this :
fprintf( file_ptr, "%l\t%ll", values.avg_get_time, values.avg_loop_time );

refered documentation

Thanks, Chicon. It turned out I was using %ull, when I needed to be using %llu. So, now, the problem is here:

         values.avg_loop_time = (unsigned long long)atoll( c );

In the above statement c is "23540235014". However, it appears that only the lower 32-bits are being converted by atoll(). I've been able to verify this with the following code:

      fgets( file_data, s, file_ptr );
      c = strtok( file_data, "\t\n" );
      values.avg_get_time = atoi( c );
      printf( "%s\n", c );
      printf( "%u\n", values.avg_get_time );
      while( c != NULL ) {
         c = strtok( NULL, "\t\n" );
         values.avg_loop_time = (unsigned long long)atoll( c );
         printf( "%s\n", c );
         printf( "%llu\n", values.avg_loop_time );
      }

Any thoughts?

Thanks.

 

[Chicon]

For the moment, I've no clue. I'm searching for well documented sites C++ programming in 64 bits. Unfortunately, I can't try the coding as I don't have a 64 bit processor.

 

[AGCurry]

I don't think C++ has a long long int, at least not in its "standard" specification.

It seems that your cast of atoll() to unsigned ought to work. You might try strtoull(). Other than that, I'm clueless (don't work with long long ints much!).

Here's a man page for strtoull().

http://developer.apple.com/documentation/Darwin/Reference/ManPages/man3/strtoull.3.html

 

[IMM]

I'm not sure what your compiler is (version-wise) or if you are using something like mingw.
I don't know if the processor you are working with is 64bit or 32 - or how you have defined the target or if you've included the headers which would resolve these issues -- but you could try using a format specifier of %I64u instead of %llu and see what it does.