[Sdcc-user] float <-> int cast

[Daniel M. <sd...@bl...>]

Dear sdcc community,
I am encountering a problem that is somehow connected to the cast
between float and int, and it is unclear to me whether I am using sdcc wrong or
whether there might be a bug somewhere. Please find below two pieces of code,
the first one works as expected, the value of debug is 6116. The second does
not work, I expect the value to be 6116, but I see a different number. It is
not clear to me whether this is always the same number and what influences its
value.

I compile for a PIC16F1938, using sdcc from svn (rev8435), configured with
--enable-new-pics (since pic16f1938 uses the "enhanced core") against gputils r932. 

$ sdcc -v
SDCC : mcs51/gbz80/z80/z180/r2k/r3ka/ds390/pic16/pic14/TININative/ds400/hc08/s08 3.2.1 #8435 (Feb 17 2013) (Linux)

I compile using
$ sdcc -I/usr/local/share/sdcc/non-free/include/ -L/usr/local/share/^Ccc/non-free/lib/pic14/ -L/usr/local/share/^Ccc/lib/pic14  -mpic14 -p16f1938 problem.c

When comparing the asm outputs I see that the first example does not call
___fs2uint, why does it not need to do this to cast the value of x? What else can I
do to narrow the problem down? Is there a debugger in which I can run these
programs on my computer and inspect the values after each step? Currently I use
a board with connected LED displays to inspect the value of debug (program
attached).

Thank you for any help.
//Daniel

$ cat problem_works.c
#define __16f1938
#include <float.h>
#include "pic14/pic16f1938.h"

static __code int __at(_CONFIG1) __CONFIG1 = _WDTE_OFF; 

unsigned int debug = 1234;

void main(void) {
	float x = 6116.0;
	debug = x;
	//debug is 6116.
} 

$ cat problem.c
#define __16f1938 
#include <float.h>
#include "pic14/pic16f1938.h" 
/* Setup chip configuration */
static __code int __at(_CONFIG1) __CONFIG1 = _WDTE_OFF; 

unsigned int debug = 1234;
float julianday = 6116.0; 

void main(void) { 
	float x = 6116.0; 

	debug = julianday;
	// expect 6116, obtain a vastly different (arbitrary?) number 
}