Re: [Jal_developers] Re: [jallist] Re: assertion failure, bug in 4.54, 4.55
- On Saturday 24 May 2003 11:11, Craig Franklin wrote:
> I removed rtllibrary.c a few days ago. It was an empty file. A commentThats ok, its not a big deal.
> said its data was moved to another file. This is an inherent problem
> with having multiple build setups.
>Its an MSVC'ism, that isn't fixible through the config scrips. Its also not a
> Not sure about the debug build. If you mean "./configure
> --enable-debug", that still works.
big deal. I'll fix it monday, I bet I broke the debug msvc compile when I
cleaned up the padantic link warnings a few days back.
> On Sat, 2003-05-24 at 08:51, Mark Gross wrote:
> > I just double checked the behavior of the Linux GCC, Cygwin-GCC and the
> > MSVC build of the the CVS tips and they all assert in codegen.c.
> > Only it now asserts at line 717.
> > Also, I see that the MSVC jal.dsp file now refrences a rtllibrary.c file
> > that doesn't exist, and the Debug build fails to link looking for
> > check_stack.
> > I'll need to fix these Monday.
> > --mgross
I don't know the reason, but the other day I compiled it as "4.54". Compiled
now as "4.55" and fails in those 2 lines.
Still working on it, but I've noticed a similar bug:
var bit y
var byte x, b, preset
x = 10
b = 5
preset = 4
b = ( preset + 5 ) - x -- Note the op_minus
> error: assertion failure in codegen.c line 787This: "y = ( preset ) >= x" compiles OK, so the bug is in statements
> file 'test2.jal' line 12 char 20
> : b = ( preset + 5 ) - x
> : -------------------^
similar to: A = ( B op C ) op D (JAL try to translate this to "D op (B op
C)", but here's a problem with op_minus, op_larger, etc.). And this is not in
"codegen" (but the error raises here), must be in "squasher" or "realloc" ... in
those files that were modified by me.
Bug is still pending.