Re: [AD] SF.net SVN: alleg:[11393] allegro/branches/4.9

[Thomas Fjellstrom <tfjellstrom@xxxxxxxxxx>]

On January 6, 2009, Thomas Fjellstrom wrote:
> On January 6, 2009, Peter Wang wrote:
> > On 2009-01-06, Thomas Fjellstrom <tfjellstrom@xxxxxxxxxx> wrote:
> > > On January 6, 2009, Evert Glebbeek wrote:
> > > > On 6-Jan-09, at 11:16 AM, Thomas Fjellstrom wrote:
> > > > >>> fix al_path_exists return type
> > > > >>
> > > > >> Shouldn't it return bool rather than int32_t?
> > > > >
> > > > > Only if booleans can store three values.
> > > >
> > > > Yes/no/maybe?
> > > > A path either exists, or it doesn't, right? Why do you need three
> > > > return values?
> > >
> > > As the code says, 1 for true, 0 for false, -1 on error.
> >
> > We also need to replace int32_ts in the path and fshook code by plain
> > int.  Using int32_t indiscriminately hides the places where a variable
> > *actually* needs to be exactly 32-bits wide, or if it's 32-bits "just
> > because" (i.e. no reason at all).
>
> All of the important places use off_t and size_t so things can deal with
> large files. Though I suppose the other int32s should be changed.. If libc
> uses plain ints, we probably should as well.
>
> For some reason I thought it was a good idea to have a consistent size of
> the types in the api. Most of them are just boolean or error codes.

One question, does Mingw change its "int" size to match up with MSVCs in 
64bit? FWIR, MSVC and GCC choose different defaults for 64bit.

> > Peter
> >
> >
> > -------------------------------------------------------------------------
> >-- ---
>
> --
> Thomas Fjellstrom
> tfjellstrom@xxxxxxxxxx
>
> ---------------------------------------------------------------------------
>---


-- 
Thomas Fjellstrom
tfjellstrom@xxxxxxxxxx