Re: [Vala] Int64 operations, I'm confused

[raum no-log org]

Ehm, ok, I've finally understood :)

I've read too fast your first answer, sorry... (and english is not my
natural langage...)

Ok, well, thanks a lot :)

Regards

Raum

> Nope, it is not strange, it is what it should be.
>
> %lld is not valid on windows, so it is taken as %d. so if you dont use
> the right format windows will execute this:
>
>  >      stdout.printf ("val6: %d, val7 : %d\n", val6, val7);
>
> instead of
>
>  >      stdout.printf ("val6: %lld, val7 : %d\n", val6, val7);
>
> What happens here is that %d will get the first 4 bytes of val6 and the
> second %d will get the next 4 bytes of val6. Resulting in 96 and 0.
> (little endian matters here).
>
> So again, please use the correct, portable format strings. %lld is not
> portable. It will work on all known unixes, but fail on windows because
> they use %I64d.
>
> On 07/31/13 10:54, raum no-log org wrote:
> > Ok, the value stored in int type variable is correct (I mean 12) but try
> > the following code :
> >
> > void main() {
> >      int64 val6 = 96;
> >      int val7 = (int) (val6 / 8);
> >      stdout.printf ("val6: %lld, val7 : %d\n", val6, val7);
> >      stdout.printf ("%d\n", val7);
> > }
> >
> > output :
> > $ a
> > val6: 96, val7 : 0
> > 12
> >
> > Int64 is correctly disdplayed with "%lld" format, int is not correctly
> > displayed at first.
> >
> > Strange ? isn'it ? :)
> >
> > But ok, I've got the right value.
> >
> >
> > generated C code :
> > /* a.c generated by valac 0.20.1, the Vala compiler
> >   * generated from a.vala, do not modify */
> >
> > #include <glib.h>
> > #include <glib-object.h>
> > #include <stdio.h>
> >
> >
> >
> >
> > void _vala_main (void);
> >
> >
> > void _vala_main (void) {
> >      gint64 val6;
> >      gint val7;
> >      FILE* _tmp0_;
> >      FILE* _tmp1_;
> >      val6 = (gint64) 96;
> >      val7 = (gint) (val6 / 8);
> >      _tmp0_ = stdout;
> >      fprintf (_tmp0_, "val6: %lld, val7 : %d\n", val6, val7);
> >      _tmp1_ = stdout;
> >      fprintf (_tmp1_, "this is correct : %d\n", val7);
> > }
> >
> >
> > int main (int argc, char ** argv) {
> >      g_type_init ();
> >      _vala_main ();
> >      return 0;
> > }
> >
> >
> >
> >
> >
> >
> > > not for me
> > >
> > > On 07/31/13 10:27, raum no-log org wrote:
> > > >       int64 val6 = 96;
> > > >         int val7 = (int) (val6 / 8);
> > >
> > > $ cat a.vala
> > > void main() {
> > >           int64 val6 = 96;
> > >           int val7 = (int) (val6 / 8);
> > >           stdout.printf ("%d\n", val7);
> > > }
> > > $ vala a.vala
> > > 12
> > >
> > > If you use the wrong format string you are going to display wrong data.
> > >
> > > This is, windows is 32bit, and values are pushed on the stack from 4
> > > bytes to 4, if the printf format expects a 64bit value it will fetch 4
> > > bytes from the value and 4 bytes more from the stack (which can be zero
> > > or any other value) causing the output you see.
> > >
> > > Please, use the proper format string.
> > >
> > > You can see the C code to verify if the compiler does it right, but you
> > > didnt provided the valac version and I doubt vala can do such a simple
> > > operation wrong.
> > >
> > > --pancake