Problem with char table defined in header file

[jyuser]

Wondering if someone could help me with this one.

I'm using PCWH compiler version 3.224 and programming at the moment PIC18F4680. I'm facing annoying problems when I define a couple of char tables in header file which I use in c file RDA interrupt function. Defining tables in header they end up blank or other way corrupted but defining them in c file just before the interrupt function they are ok.

Here is some c file code (pay attention to char tables and their test printing):

char set_voltage_command[11] = {'s', 'e', 't', ' ', 'v', 'o', 'l', ' ', 'c', 'h'};
char set_polarity_command[11] = {'s', 'e', 't', ' ', 'p', 'o', 'l', ' ', 'c', 'h'};


#int_RDA
RDA_isr()
{
int8 command_string_length = 0;
int8 new_polarity = 0;
char channel_char = '0';
int8 channel_numb = 0;
int8 go = 1;
int8 i = 0, j = 0;
int8 new_voltage = 0;
int8 k = 0; // Needed for testing at the end

command_string_length = my_get_string(rs232_input_buffer, BUFFER_SIZE);
//printf("Command string length: %u\n\n", command_string_length); // For testing
printf("\nHelp: %s\n", help_command); // For testing
for(k = 0; k < 4; k++) {
printf("%c", help_command[k]);
}
printf("\nGet all: %s\n", get_command); // For testing
for(k = 0; k < 10; k++) {
printf("%c", get_command[k]);
}
printf("\nSet pol: %s\n", set_polarity_command); // For testing
for(k = 0; k < 10; k++) {
printf("%c", set_polarity_command[k]);
}
printf("\nSet vol: %s\n", set_voltage_command); // For testing
for(k = 0; k < 10; k++) {
printf("%c", set_voltage_command[k]);
}

And some h file code:

char help_command[] = {'h', 'e', 'l', 'p', '0'};
char get_command[] = {'g', 'e', 't', ' ', 'a', 'l', 'l', ' ', 'c', 'h', '0'};

//char help_command[] = {'h', 'e', 'l', 'p'};
//char get_command[] = {'g', 'e', 't', ' ', 'a', 'l', 'l', ' ', 'c', 'h'};

//char set_voltage_command[] = {'s', 'e', 't', ' ', 'v', 'o', 'l', ' ', 'c', 'h'};
//char set_polarity_command[] = {'s', 'e', 't', ' ', 'p', 'o', 'l', ' ', 'c', 'h'};


So in one line: c file definitions work, h file definitions don't.

[Mike Walne]

I think I know what you are telling us.

To avoid possible crossed lines, it would be easier if you post complete code which is stripped to the MINIMUM, so we can compile as is.

Mike

[andyfraser]

This is also a shocking example of how to use the RDA interrupt.
It needs to be short and quick you should not be getting a complete string nor should you be calling printf from within this interrupt.
Look at the examples provided by CCS on receiving strings using a circular buffer under interrupt.

Andy

[ckielstra]

Several comments:
- You didn't post the test print results.
- When posting code please use the 'code' tags to preserve the formatting for easier reading.

[RF_Developer]

...And '0' is the character zero, not NULL or the "zero terminator", which is '\0'.

RF Developer.

[jyuser]

Thanks for all the helpful comments! I really could correct some of my bad programming habits. Circular buffer seems to be the way to do it...

One more questions came up: when I handle received string by polling in main method, it appears that there is some delay between the first sent char and the rest of the string in terms of arrival (to the buffer).

Here is a simple example:
sending from PC via RS-232 a command 'get voltage' will appear in buffer first only as 'g' if constantly polled in program. Second time checking the buffer there will be the rest of the command 'et voltage'. Solution is adding some delay to polling, but do you know is there really some HW delay after first handled char?