problems compiling pforth

[Dwight Schauer]

From: "Phil Burk" 
To: "Dwight Schauer" 
Date: Sun, 16 Feb 2003 19:52:37 -0800

Hello Dwight,

Thanks for the notes. I'll apply the mods for the next rev.

ALso you may want to download the "pfthpc21.zip". The project is for Windows
but the portable source code will work with Unix and is more recent.

Sorry that I haven't updated  the UNix rev in a while. Other projects demand
my time.

Phil Burk

----- Original Message -----
From: "Dwight Schauer" 
To: 
Sent: Sunday, February 16, 2003 7:19 PM
Subject: pfthunix19
> Hello Phil,
>
> I just downloaded pfthunix19.tar.Z
>
> I compiled it with gcc (GCC) 3.2.1 20021207 (Gentoo Linux 3.2.1-20021207)

and
> ran into a few problems. I thought I'd let you know, as you set you wanted

to
> be notified of such things.
>
> I ran into some problems and had to make a few changes.
>
> 20:22 < Stepan> csrc/pfcustom.c:85: invalid conversion from `int32

(*)(long
> int)' to `void*'
> 20:22 < Stepan> csrc/pfcustom.c:85: invalid conversion from `void (*)(long
> int, long int)'
>
> --- pfcustom.c.orig 2003-02-16 20:48:48.000000000 -0600
> +++ pfcustom.c 2003-02-16 20:49:35.000000000 -0600
> @@ -80,8 +80,8 @@
>
> void *CustomFunctionTable[] =
>  {
> CTest0,
> CTest1
> (void*)CTest0,
> (void*)CTest1
>
> };
>  #endif
>
> 20:49 < Dwight> objects/pf_cglue.o(.eh_frame+0x11): undefined reference to
>                 `__gxx_personality_v0'
> 20:49 < Dwight> objects/pf_core.o(.eh_frame+0x11): undefined reference to
>                 `__gxx_personality_v0'
> 20:49 < Dwight> objects/pf_inner.o(.eh_frame+0x11): undefined reference to
>                 `__gxx_personality_v0'
> 20:49 < Dwight> objects/pf_main.o(.eh_frame+0x11): undefined reference to
>                 `__gxx_personality_v0'
> 20:49 < Dwight> objects/pf_save.o(.eh_frame+0x11): undefined reference to
>                 `__gxx_personality_v0'
> 20:49 < Dwight> objects/pf_text.o(.eh_frame+0x11): more undefined

references
> to
>                 `__gxx_personality_v0' follow
> 20:49 < Dwight> collect2: ld returned 1 exit status
> 20:49 < Dwight> make: *** [pforth] Error 1
>
> --- makefile.orig 1998-04-06 19:43:01.000000000 -0500
> +++ makefile 2003-02-16 20:52:18.000000000 -0600
> @@ -27,7 +27,7 @@
>          -Wmissing-declarations
>
> CCOPTS = -DPF_SUPPORT_FP $(FULL_WARNINGS)
> -COMPILER = gcc
> +COMPILER = g++
>
> EMBCCOPTS = -DPF_NO_INIT -DPF_NO_MALLOC -DPF_NO_FILEIO \
>      -DPF_NO_CLIB -DPF_STATIC_DIC
>
> pforth -i system.fth
> make: pforth: Command not found
> make: *** [pforth.dic] Error 127
>
> To make, I had to do a "export PATH=$PATH:. make", as pforth was being
> executed while not in the path.
>
> Dwight

_______________________________________________
OpenBIOS mailing list
[e-mail]
http://freiburg.linux.de/cgi-bin/mailman/listinfo/openbios

[Stefan Reinauer]

* Dwight Schauer  [030217 11:06]:
> > 20:49 < Dwight> objects/pf_cglue.o(.eh_frame+0x11): undefined reference to
> >                 `__gxx_personality_v0'

[..]
> > --- makefile.orig 1998-04-06 19:43:01.000000000 -0500
> > +++ makefile 2003-02-16 20:52:18.000000000 -0600
> > @@ -27,7 +27,7 @@
> >          -Wmissing-declarations
> >
> > CCOPTS = -DPF_SUPPORT_FP $(FULL_WARNINGS)
> > -COMPILER = gcc
> > +COMPILER = g++

It might be worth to just not use C++ at all:

--- makefile.orig       1998-04-07 02:43:01.000000000 +0200
+++ makefile    2003-02-17 13:27:53.000000000 +0100
@@ -18,7 +18,6 @@
         -fno-unroll-loops \
         -fpeephole \
         -fno-keep-inline-functions \
-        -x c++ \
         -Wcast-qual \
         -Wall \
         -Wwrite-strings \

With above patch, pforth is 30k instead of 34k, doing the same thing.
Is there any reason this was originally there?

Stefan

[Dwight Schauer]

On Monday 17 February 2003 06:45 am, Stefan Reinauer wrote:
> It might be worth to just not use C++ at all:
>
> --- makefile.orig       1998-04-07 02:43:01.000000000 +0200
> +++ makefile    2003-02-17 13:27:53.000000000 +0100

[...]
>    -x c++ \

[...]
> With above patch, pforth is 30k instead of 34k, doing the same thing.
> Is there any reason this was originally there?

I don't know why. I did'nt investigate it. I just was trying to make it 
compile. I agree, it should'nt be c++, especially if it compiles fine as C 
code.

Your patch is better, I was just going for quick and dirty to see what was 
there.

Dwight

_______________________________________________
OpenBIOS mailing list
[e-mail]
http://freiburg.linux.de/cgi-bin/mailman/listinfo/openbios