Re: tests

Date view Thread view Subject view Author view

From: Godmar Back (gback@cs.utah.edu)
Date: Wed Feb 03 1999 - 15:12:36 EST


>
> On Feb 3, 1999, Godmar Back <gback@cs.utah.edu> wrote:
>
> > As I understand it, the types get pulled in by seemingly harmless
> > includes such as <stdio.h> or <stdlib.h>.
>
> Except that they're highly system dependent (i.e., most systems don't
> define it), so we'd better not rely on them.
>

"They" meaning uint32 etc.?

So you're proposing the change all uses in Kaffe to, say kaffe_uint32
or jint or some such?

That's fine with me, actually. I didn't want that because it would
involve lots of changes.

> > So the question becomes how to avoid double definitions introduced by
> > gtypes.h.
>
> We shouldn't define them, unless configure determines that they are
> not defined. And even then there's risk for double definition, so
> we'd better avoid it.
>
> > What do you mean by "relies on those variables instead of using jtypes"?
>
> I mean ``what the hack is that guy doing with int8_t et al that he
> couldn't do with jbyte, jshort, jint and jlong''?
>

It's not him, Alexandre.
The non-beos parts of Kaffe rely on them all over the place.
He just hit the problem cause beos defined them.

I am all for eliminating those dependencies.

In fact, I am against typedefs for basic types in general. Whenever you
don't need to specify a range or signedness, you should use what C gives you.

        - Godmar


Date view Thread view Subject view Author view

This archive was generated by hypermail 2b29 : Sat Sep 23 2000 - 19:57:57 EDT