RPG exec sql works only once - sql

I wrote program in RPG IV free-form and i met a problem. Im trying to execute sql statment with:
dcl-proc getRecord;
dcl-pi *N int(5) end-pi;
dcl-s rec int(5) inz(0);
exec sql
select max(SCORE) into :rec from SNK_HISC;
return rec;
END-PROC;
It works only once after compilation. Maybe i should reset some indicators or sql flags? At the end of program i use:
*inlr = *on;
return;
Where procedure is called:
exfmt BACK;
exfmt INFO;
inital();
dow run and alive;
write MAINBOARD;
READ(E) AKE_DSP;
run = FKeyListener();
alive = goForward();
ENDDO;
LSCORE = points;
if points > getRecord();
LRECORD = 'NEW RECORD!';
exec sql
insert into AKE_HISC( USID , SCORE) VALUES( :userID , :points);
ENDIF;
exfmt ENDING;
*inlr = *on;
return;
And in this procedure:
dcl-proc inital;
snake(1).x = 5;
snake(2).x = 5;
snake(3).x = 5;
snake(1).y = 5;
snake(2).y = 6;
snake(3).y = 7;
direction.x = 1;
direction.y = 0;
SCORE = points;
RECORD = getRecord();
genSnack();
refresh();
END-PROC;
I run program with this CL program:
PGM
MONMSG MSGID(CPF0000 MCH0000) EXEC(GOTO CMDLBL(END))
OVRDSPF FILE(AKE_DSP) DEV(*REQUESTER) WAITRCD(1)
CALL PGM(AKE_S)
END:
DLTOVR FILE(*ALL)
MONMSG MSGID(CPF0000)
ENDPGM

Problem was not with sql query but with that where it is. I removed procedure getRecord() and make rec global variable and the sql query was placed inside the main cycle thats all.

Related

Verilog HDL syntax error at test_bench_lb2.v(14) near text "genvar"; expecting "end"

I am new one in Verilog and I need to write a simple test bench, but I get an error and I cannot understand why it is
Here is my code
`timescale 1 ns / 1 ns
module test_bench_lb2;
reg [12:0] in_lines_tb;
wire [4:0] out_lines_tb;
wire error_tb;
localparam PERIOD = 10;
initial
begin
genvar i;
for(i = 0; i <= 8000; i = i + 1)
begin
in_lines_tb = i;
#PERIOD;
end
#(PERIOD*20) $stop;
end
initial
begin
$monitor("time = %time in_lines = %b out_lines = %b error = %b",
$time, in_lines_tb, out_lines_tb, error_tb);
end
DESHIFRATOR inst1(.in_lines(in_lines_tb), .out_lines(out_lines_tb), .error(error_tb));
endmodule
This
genvar i;
should be this
integer i;
A genvar is a special kind of variable used in a construct called a generate loop. You code uses an ordinary for loop.

Passing a signal name into a verilog task

Instead of this task just toggling tb.stimulus.top.Ichip0.vbiash high and low ten times I would like to be able to call it passing in any signal tb.stimulus.top.Ichip0.vbiasl, tb.stimulus.top.Ichip0.vbiasx, or tb.stimulus.top.Ichip0.vbiasz and make them toggle as well. For example toggle_signal(tb.stimulus.top.Ichip0.vbiasl); Is it possible to do this. If so I would really appreciate an example of how I would accomplish this.
task toggle_signal;
begin
for (monpad_index=0; monpad_index < 10; monpad_index = monpad_index + 1)
begin
#1000;
force tb.stimulus.top.Ichip0.vbiash = 1'b1;
#1000;
force tb.stimulus.top.Ichip0.vbiash = 1'b1;
#1000;
end
end
You cannot pas the name of a single to a task. But you can create a macro to do this.
`define toggle_signal(sig) \
for (monpad_index=0; monpad_index < 10; monpad_index = monpad_index + 1) \
begin \
#1000 force tb.stimulus.top.Ichip0.sig = 1'b1; \
#1000 force tb.stimulus.top.Ichip0.sig = 1'b0; \
#1000; \
end
and then write
`toggle_signal(vbiash)
`toggle_signal(vbiasl)
If you are not afraid do dive into a little C programming, you could write your own PLI/VPI. I suggest checking out IEEE Std 1800-2012's sections on PLI/VPI (§ 36, 37, & 38). Your come may looks something like the following (Note, the code is a starting reference. I haven't tested it):
static int my_vpi_force_release_calltf(PLI_BYTE* user_data) {
vpiHandle sys, argv, signal;
p_vpi_value p_value;
int force_release;
force_release = (int) user_data;
sys = vpi_handle(vpiSysTfCall, 0);
argv = vpi_iterate(vipArgument, sys);
signal = vpi_handle_by_name(vpi_get_str(vpi_scan(argv), 0);
if (force_release != 0 ) {
vpi_get_value(vpi_scan(argv), p_value);
vpi_put_value(signal, p_value, 0, vpiForceFlag);
} else {
vpi_get_value(signal, p_value);
vpi_put_value(signal, p_value, 0, vpiReleaseFlag);
}
return 0;
}
void register_my_vpi_force_release()
{
s_vpi_systf_data data;
data.type = vpiSysTask;
data.calltf = my_vpi_force_release_calltf;
data.tfname = "$my_force";
data.user_data = (PLI_BYTE8 *) 1;
vpi_register_systf(&data);
data.tfname = "$my_release";
data.user_data = (PLI_BYTE8 *) 0;
vpi_register_systf(&data);
}
You can call your PLI/VPI tasks in verilog:
task toggle_signal( input [80*8-1:0] path_str );
integer index;
begin
for (index=0; index < 10; index = index + 1)
begin
#1000;
$my_force( path_str, 1'b1);
#1000;
$my_force( path_str, 1'b0 );
#1000;
end
$my_release( path_str );
end
endtask
...
initial begin
...
toggle_signal( "tb.stimulus.top.Ichip0.vbiash" );
...
end
If you don't want to write your own custom PLI/VPI, then I suggest enabling SystemVerilog and include UVM (the major simulators have UVM build in, or download it yourself). The UVM library has a build in method uvm_hdl_force/uvm_hdl_release.
Here you go. The following code passes a signal name as a parameter to a task then prints the signal name as a string. Tested in iverilog.
reg this_is_some_signal_name;
`define NAME(net) `"net`"
task print;
input reg [32*8] str;
begin
$display(">>> %0s", str);
end
endtask
initial begin
print(`NAME(this_is_some_signal_name));
$finish;
end
Outputs:
>>> this_is_some_signal_name

Verilog module output reg driving output reg?

So I'm trying to instantiate a module within a module. The root module has output ports that drive the output pins and I want the inner module to drive those ports directly but I can't get it work anyway round.
/*
A root module for the 005_135-scanner_mainboard_revA_sch-pcb. Can be used as a test bench for testing peripheral devices but has been designed to be the master root module for the final code.
All IO is included and reset pins on peripheral devices driven active reset with data lines driven to an appropriate value (located in the ‘initial block’).
George Waller. 09/08/15.
*/
module root_module( ft_reset, ft_usb_prsnt, ft_bus_pwrsav, ft_bus_oe, ft_bus_clkout, ft_bus_siwu, ft_bus_wr, ft_bus_rd, ft_bus_rxf, ft_bus_txe, ft_bus_d,
mtr_fault, mtr_config, mtr_m1, mtr_m0, mtr_rst, mtr_out_en, mtr_step, mtr_dir,
ccd_driver_oe, ccd_p1, ccd_p2, ccd_cp, ccd_rs, ccd_sh,
dac1_sdin, dac1_sclk, dac1_sync, dac2_sdin, dac2_sclk, dac2_sync,
adc_pwrdn, adc_encode, adc_d,
fpga_reset,
clk,
gpio,
led_ctrl,
leds);
//Input declarations
input wire ft_usb_prsnt, ft_bus_clkout, ft_bus_rxf, ft_bus_txe,
mtr_fault,
fpga_reset,
clk;
input wire [7:0] adc_d;
//Output declarations
output reg ft_reset, ft_bus_pwrsav, ft_bus_oe, ft_bus_siwu, ft_bus_wr, ft_bus_rd,
mtr_config, mtr_m1, mtr_m0, mtr_rst, mtr_out_en, mtr_step, mtr_dir,
ccd_driver_oe, ccd_p1, ccd_p2, ccd_cp, ccd_rs, ccd_sh,
adc_pwrdn, adc_encode,
led_ctrl;
output reg dac1_sdin, dac1_sclk, dac1_sync, dac2_sdin, dac2_sclk, dac2_sync;
output reg [7:0] leds;
//Input output declarations.
inout reg [7:0] ft_bus_d;
inout reg [16:0] gpio;
//Variables go here
integer count, count1, state, pixel_n, line_n, t_int;
integer data[8];
reg en;
//Initial values on start up.
initial
begin
//IO initial setup values.
ft_reset = 1; ft_bus_pwrsav = 1; ft_bus_oe = 1; ft_bus_siwu = 0; ft_bus_wr = 1; ft_bus_rd = 1; //NEED TO APPLY REAL VAULES!!!
mtr_config = 1; mtr_m1 = 1; mtr_m0 = 1; mtr_rst = 1; mtr_out_en = 1; mtr_step = 0; mtr_dir = 0;
ccd_driver_oe = 1; ccd_p1 = 0; ccd_p2 = 0; ccd_cp = 0; ccd_rs = 0; ccd_sh = 0;
dac1_sdin = 0; dac1_sclk = 0; dac1_sync = 0; dac2_sdin = 0; dac2_sclk = 0; dac2_sync = 0;
adc_pwrdn = 0; adc_encode = 0;
led_ctrl = 0;
leds = 0;
gpio = 0;
ft_bus_d = 0;
//Variables setup values.
count = 0;
count1 = 0;
state = 0;
pixel_n = 0;
line_n = 0;
t_int = 10000; //t_int = integration time. integration time (seconds) = t_int * 10x10^-9.
end //End initial
//Some other code goes here.
always #(posedge ft_bus_clkout)
begin
if(count == 50000000)
begin
en <= 1;
count = 0;
end
else
begin
en <= 0;
count = count + 1;
end
end //End always
AD5601_module AD5601(.en(en), .clk(clk), .data(127),.sdout(dac1_sdin),
.sclk(dac1_sclk), .sync(dac1_sync));
endmodule //End module.
And the inner module:
module AD5601_module(en, clk, data, sdout, sclk, sync);
input wire clk;
input wire en;
input wire [7:0] data;
output reg sdout, sclk, sync;
integer sclk_count;
integer data_state;
integer delay_counter;
integer pd[2];
initial
begin
sclk_count = 0;
data_state = 99;
delay_counter = 0;
pd[0] = 0;
pd[1] = 0;
sdout = 0;
sclk = 0;
sync = 1;
end
always # (posedge en)
begin
if(data_state == 99)data_state <= 0;
end
always # (posedge clk)
begin
if(sclk_count == 49)
begin
sclk_count = 0;
sclk = ~sclk;
end
else sclk_count = sclk_count + 1;
end
always # (posedge sclk)
begin
case(data_state)
0: begin
sync = 0;
sdout <= pd[1];
data_state <= 1;
end
1: begin
sdout <= pd[0];
data_state <= 2;
end
2: begin
sdout <= data[7];
data_state <= 3;
end
3: begin
sdout <= data[6];
data_state <= 4;
end
4: begin
sdout <= data[5];
data_state <= 5;
end
5: begin
sdout <= data[4];
data_state <= 6;
end
6: begin
sdout <= data[3];
data_state <= 7;
end
7: begin
sdout <= data[2];
data_state <= 8;
end
8: begin
sdout <= data[1];
data_state <= 9;
end
10:begin
sdout <= 0;
if(delay_counter == 6)
begin
data_state <= 99;
delay_counter <= 0;
sync = 1;
end
else delay_counter = delay_counter + 1;
end
endcase
end
endmodule
So with the code as is I get the error
'output or inout port should be connected to a structural net
expression'
.
If I change the outputs in the root module or the inner module to wire I get the error 'the expression on the left hand side should be of a variable type'.
So at this point I have no idea how you nest outputs! Some help would be appreciated!
Thanks
inout ports should be a net type (wire or tri) and not a reg. A reg does not have conflict resolution (when there are two or more active driver). An inout should not be assigned in a procedural block (e.g. always-block, initial-block). It should be a simple assign statement as below. The designer must ensure there is only on active driver on the IO at any point.
assign io_port_name = driver_enable ? io_out_value : 'bz; // io_out_value should be a flop
An output should only be declared an output reg if it is assigned in a procedural block (e.g. always-block, initial-block) within the current module. All other outputs should be output or output wire (these identifiers are synonymous; the former is implicit while the latter is explicit). An should only be assigned within one always. FPGAs allow initial blocks, ASIC/IC does not.
If SystemVerilog is enabled, replace output reg with output logic. logic can be used for flops and single directional nets. logic is not recommended for inout; logic like reg does not have conflict resolution.
The arrays integer data[8]; and integer pd[2]; are SystemVerilog syntax and not compatible with Verilog. Either enable SystemVerilog or change to integer data[0:7]; and integer pd[0:1];
SystemVerilog can be enabled per file by changing the file extension from .v to .sv; recommended. Simulators/synthesizers typically have a switch to force all Verilog to be treated as SystemVerilog; not recommended, refer to the simulator/synthesizer manual if desired to go this route.

Conditional code in Verilog in a generate loop

I am trying to encode conditional behavior for Verilog statements in a generate loop. For example, the code below returns an error.
module <something> (out);
parameter [4:0] someParam = 0;
output [5:0] out;
genvar l, m;
for(l=0; l<5; l=l+1) begin:STAGE_1
m = 0;
if(someParam[l] < 2)
m = l+2;
else begin
m = l-2;
end
if (m>16) assign out[l] = 1'b0;
else assign out[l] = 1'b1;
end
endmodule
The problem is that the variable m is not a constant and the code errors out. Is there any way I can use compile time variable inside a generate statement which would allow some functionality like the variable m above?
Thanks.
I didnt understand what you intended to calculate due to some errors in your code.
In general, for you to use a parameter in a statement you can use an always block with a if statement as following:
module <something> (out);
parameter [4:0] someParam = 0;
output out; // in this case out is only one bit. it can be more of course.
integer l,m; // no need for genver when not using generate
always (*) begin
m = 0;
for (l=0; l<5; l=l+1) begin:STAGE_1
if (someParam[l] == 1'b1) // nothing good comes for checking if a bit is less then 2
m = m+1; // just counting bits in someParam. doing +-2 does not make sense.
end
if (m >= 3)
out = 1'b1;
else
out = 1'b0;
end
The above is a majority function.
Good luck

Oracle Pro*C updating table with cursor failed

I have a table like this:
CREATE TABLE book_info (
book_id VARCHAR(32) not null,
title varchar(255) not null,
author varchar(255) not null,
folder_path varchar(255) not null,
primary key(book_id)
);
And i insert this data on it:
insert into book_info values('BOOK1', 'APUE', 'Richard Stevens', '/home/user1/unix_programming_books');
insert into book_info values('BOOK2', 'Unix Network programming', 'Richard Stevens', '/home/user1/unix_programming_books');
insert into book_info values('BOOK3', 'Core Python Applications Programming', 'Wesley J. Chun', '/home/user1/python_programming_books');
I'm trying to update this table using Oracle PRO*C, but i can't! below is my code:
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
EXEC SQL INCLUDE SQLCA;
EXEC SQL INCLUDE ORACA;
#define USER_LEN 10
#define PASS_LEN 10
VARCHAR user[USER_LEN];
VARCHAR pass[PASS_LEN];
#define STRCPY_TO_ORA(dest, source)\
dest.len = strlen(source);\
strcpy((char *)dest.arr, (const char *)source)
#define STRCPY_FROM_ORA(dest, source)\
source.arr[source.len] = 0;\
strcpy((char *)dest,(const char *)source.arr)
/* Connecting to the database */
int db_connection(char *db_user, char *db_pass)
{
strncpy((char *) user.arr, db_user, USER_LEN);
user.len = strlen((char *) user.arr);
strncpy((char *) pass.arr, db_pass, PASS_LEN);
pass.len = strlen((char *) pass.arr);
EXEC SQL CONNECT :user IDENTIFIED BY :pass;
if (sqlca.sqlcode != 0)
{
fprintf(stdout, "Connection failed:%s\n", sqlca.sqlerrm.sqlerrmc);
return(sqlca.sqlcode);
}
fprintf(stdout, "Connected to ORACLE as user:%s\n", user.arr);
return (sqlca.sqlcode);
}
int book_not_found_function(char *path)
{
fprintf(stdout, "%s\n", __FUNCTION__);
}
int path_update_success_function(char *book_id, char *new_path)
{
fprintf(stdout, "Update book %s path to %s\n", book_id, new_path);
}
void other_function(void)
{
fprintf(stdout, "%s\n", __FUNCTION__);
}
/* Updating books path */
int books_path_updating(char *old_path, char *new_path)
{
char book_id_string[32];
EXEC SQL BEGIN DECLARE SECTION;
varchar sql_old_path[255];
varchar sql_new_path[255];
varchar sql_book_id[32];
EXEC SQL END DECLARE SECTION;
STRCPY_TO_ORA(sql_old_path, old_path);
STRCPY_TO_ORA(sql_new_path, new_path);
/* Declare a cursor for the FETCH statement. */
EXEC SQL DECLARE books_cursor CURSOR FOR
SELECT BOOK_ID
FROM BOOK_INFO
WHERE FOLDER_PATH = :sql_old_path;
if (sqlca.sqlcode != 0)
{
fprintf(stdout, "Declare cursor failed\n");
fprintf(stdout, "Oracle error %s\n", sqlca.sqlerrm.sqlerrmc);
return(sqlca.sqlcode);
}
EXEC SQL OPEN books_cursor;
if (sqlca.sqlcode != 0)
{
fprintf(stdout, "Open cursor failed\n");
fprintf(stdout, "Oracle error %s\n", sqlca.sqlerrm.sqlerrmc);
return(sqlca.sqlcode);
}
for ( ;; )
{
//EXEC SQL WHENEVER NOT FOUND DO break; // I used it but still nothing
//EXEC SQL WHENEVER NOT FOUND GOTO not_found; // I used this too
//EXEC SQL WHENEVER NOT FOUND DO continue; // I used this too
/* Fetching data */
EXEC SQL FETCH books_cursor
INTO :sql_book_id;
if (sqlca.sqlcode == 1403)
{
fprintf(stdout, "No book found for this folder %s\n", old_path);
book_not_found_function(old_path);
return 0;
}
else if (sqlca.sqlcode != 0)
{
fprintf(stdout, "Oracle error %s\n", sqlca.sqlerrm.sqlerrmc);
EXEC SQL CLOSE books_cursor;
return (sqlca.sqlcode);
}
else
{
STRCPY_FROM_ORA(book_id_string, sql_book_id);
fprintf(stdout, "BOOK_ID = %s\n", book_id_string);
/* Updating the path */
EXEC SQL UPDATE BOOK_INFO
SET FOLDER_PATH =:sql_new_path
WHERE BOOK_ID =:sql_book_id;
if (sqlca.sqlcode != 0)
{
fprintf(stdout, "Oracle error %s\n", sqlca.sqlerrm.sqlerrmc);
EXEC SQL CLOSE books_cursor;
return (sqlca.sqlcode);
}
else
{
path_update_success_function(book_id_string, new_path);
}
}
}
EXEC SQL CLOSE books_cursor;
other_function();
EXEC SQL COMMIT WORK RELEASE;
return 0;
}
int main(int argc, char **argv)
{
db_connection("evariste", "123456");
books_path_updating("/home/user1/unix_programming_books", "/home/user1/UNIX_PROGRAMMING_BOOKS");
books_path_updating("/non_existing_path", "/non_existing_path");
return 0;
}
This program produce the output :
Connected to ORACLE as user:evariste
BOOK_ID = BOOK1
Update book BOOK1 path to /home/user1/UNIX_PROGRAMMING_BOOKS
BOOK_ID = BOOK2
Update book BOOK2 path to /home/user1/UNIX_PROGRAMMING_BOOKS
No book found for this folder /home/user1/unix_programming_books // WHEY THIS?
book_not_found_function // WHY THIS
Declare cursor failed // WHY THIS
Oracle error ORA-01403: no data found // WHY THIS
The table is not updated and the functions path_update_success_function and other_function are never executed! Why this?
Thanks for your help.
Connected to ORACLE as user:evariste
BOOK_ID = BOOK1
Update book BOOK1 path to /home/user1/UNIX_PROGRAMMING_BOOKS
BOOK_ID = BOOK2
Update book BOOK2 path to /home/user1/UNIX_PROGRAMMING_BOOKS
No book found for this folder /home/user1/unix_programming_books // WHEY THIS?
You've fetched past the end of the resultset. Two rows match the cursor this time, so the first two fetches succeed; the third gets no data. This is expected and you shouldn't treat this as an error - just break the loop, which will also cause other_function to be called.
book_not_found_function // WHY THIS
Because you treat 1403 as an error. If you want to call this function when there are no matches, you'll need to count in the loop, and call it afterwards if needed.
Declare cursor failed   // WHY THIS
Oracle error ORA-01403: no data found // WHY THIS
sqlca.sqlcode seems to still be set from the earlier fetch, so this is misleading.
As far as I can remember you'd normally declare the cursor once in the file, not each time the function is called; not sure if Pro*C just ignores the redefinition. You can look at the generated file to see how it deals with it. You also won't get a runtime error from this, if it's wrong it won't (pre-)compile.
The table is not updated and the functions
path_update_success_function and other_function are never executed!
Why this?
path_update_success is called for the first two fetches, but not for the third which fails, and not for the second path because the function returns due to the apparent declare cursor before it gets near it. other_function isn't called because for both calls you return from the function before you can reach it. Similarly the table seems to not be updated because you return before you commit. Unlike SQL*Plus, Pro*C doesn't automatically commit on exit, so there is an implicit rollback. Note also that if you did get to the commit, the release disconnects you, so the second time you'd get a not-connected-to-Oracle error. You should decide to commit/rollback once really, probably right at the end of main.
Untested modification:
int books_path_updating(char *old_path, char *new_path)
{
char book_id_string[32];
int books_found;
EXEC SQL BEGIN DECLARE SECTION;
varchar sql_old_path[255];
varchar sql_new_path[255];
varchar sql_book_id[32];
EXEC SQL END DECLARE SECTION;
STRCPY_TO_ORA(sql_old_path, old_path);
STRCPY_TO_ORA(sql_new_path, new_path);
/* Declare a cursor for the FETCH statement */
EXEC SQL DECLARE books_cursor CURSOR FOR
SELECT BOOK_ID
FROM BOOK_INFO
WHERE FOLDER_PATH = :sql_old_path;
EXEC SQL OPEN books_cursor;
if (sqlca.sqlcode != 0)
{
fprintf(stdout, "Open cursor failed\n");
}
books_found = 0;
while (sqlca.sqlcode == 0)
{
/* Fetching data */
EXEC SQL FETCH books_cursor
INTO :sql_book_id;
if (sqlca.sqlcode != 0)
{
break;
}
STRCPY_FROM_ORA(book_id_string, sql_book_id);
fprintf(stdout, "BOOK_ID = %s\n", book_id_string);
/* Updating the path */
EXEC SQL UPDATE BOOK_INFO
SET FOLDER_PATH = :sql_new_path
WHERE BOOK_ID = :sql_book_id;
if (sqlca.sqlcode != 0)
{
break;
}
/* Track how many books we found, though we only really care later that
* this is non-zero */
books_found++;
path_update_success_function(book_id_string, new_path);
}
EXEC SQL CLOSE books_cursor;
/* Check for and display error, but ignore 1403 as this just indicates the
* end of the result set */
if ((sqlca.sqlcode != 0) && (sqlca.sqlcode != 1403))
{
fprintf(stdout, "Oracle error %s\n", sqlca.sqlerrm.sqlerrmc);
return 1;
}
if (books_found == 0)
{
fprintf(stdout, "No book found for this folder %s\n", old_path);
book_not_found_function(old_path);
return 0;
}
other_function();
return 0;
}
int main(int argc, char **argv)
{
int rc;
rc = db_connection("evariste", "123456");
/* Only do the first path if we didn't get an error connecting */
if (rc == 0)
{
rc == books_path_updating("/home/user1/unix_programming_books",
"/home/user1/UNIX_PROGRAMMING_BOOKS");
}
/* Only do the next path if we didn't get an error from the previous one */
if (rc == 0)
{
rc = books_path_updating("/non_existing_path",
"/non_existing_path");
}
/* Decide whether to rollback or commit; this assumes you don't want to
* keep any changes if there are any errors */
if (rc != 0)
{
EXEC SQL ROLLBACK WORK RELEASE;
return 1;
}
EXEC SQL COMMIT WORK RELEASE;
return 0;
}
Seems DECLARE CURSOR does not set sqlca.sqlcode.
The error you've got was the previous one when you achieved the end of your fetch.
So you should not test errors after a DECLARE CURSOR, but after the OPEN CURSOR only.