Re: test rig first cut (list discussion, practice etc)

[Remco Treffkorn <remco@rvt.com>]

On Friday 11 July 2003 01:19, Ethan Benson wrote:
> On Thu, Jul 10, 2003 at 01:42:25PM -0700, Remco Treffkorn wrote:
...
> yes, and i think you for your help, i just ask you please read my
> replies a bit more carefully, as ive had to say the same thing over
> and over.
>
> > Eathan, I have asked questions. If some of the answers are obvious to
> > you,
>
> spelling.

ditto.

> > I have not made any proposals yet. I have described how I think things
> > ARE. I
>
> you keep making statements that the vfs/filesystem stuff will not be
> in prom-libc.  im telling you it WILL.

Semantics. Prom-libc sits on top of a layer that implements vfs that sits on 
top of the prom. None of libc has to be aware of the vfs stuff. This is a 
statement you may disagree with, but it is really a question about my basic 
understanding of your concept.

>
> > may very well be wrong about some aspects. That's when you start
> > teaching... I need to fully understand the status quo, and I need to
> > understand where you want to go from here. If there is a yaboot2 white
> > paper, that would be a good start. As it is, you seem to be the single
> > source of information.
>
> your tarball wires read/write etc directly to linux syscalls, this is
> WRONG, as it will bypass the vfs layer in prom-libc, not allowing it
> to be tested.  linux syscalls will be __linux_open() etc.  only rawio
> will call them.

What is there makes a test program compile. It does not make it correct. Give 
me a few more days to finish it.

...
> as i said above my current mood has nothing to do you with you, but
> some of my conversations with you have been frustrating as its seemed
> like you were not listening.

ditto.

...
> shouldn't take any asm stuff.. i don't have any of these vaargs problems..
>
> and i have done a basic test under gcc 3.3 in debian, things worked
> fine (via the LD_PRELOAD hack test).  i think your compiler is
> broken.  (its -dumpmachine output certainly is).

My compiler comes with YellowDog Linux 3.0.
I am not in the habit of compiling gcc myself. I operate under the ssumption, 
that if a major distribution (terra soft loves to hear that) is broken, they 
just created a new standard. Many people are using this buggy compiler. You 
can either accommodate them, or be innundated by e-mails complainig about 
your package being broken. And I know you will love this.

Again, I am not suggesting you change anything. I have worked around this bug 
without complaining, to either you, or terra soft.

Let's talk about varargs in the appropriate email.

-- 
Remco Treffkorn (RT445)
HAM DC2XT
remco@rvt.com   (831) 685-1201