Subject:
|
Re: gcc build instructions for linux
|
Newsgroups:
|
lugnet.robotics.rcx.legos
|
Date:
|
Thu, 8 Apr 1999 04:15:57 GMT
|
Viewed:
|
1490 times
|
| |
|
|
OK, I have some more information now. When I build 0.1.6 in my new environment
it works fine. 0.1.7 and the snap exhibit the problem.
The problem is that when using tm-and-ir, the 'view' button (which is supposed
to send the message "Hello World") causes the IR port on the RCX to go
always-on (at least that's what it looks like with my Radio Shack IR
detector card) and of course does not send the message. The RCX will no
longer display received characters after this, though it does stop running
when pressing the 'run' button (though this does not turn off the IR). Pressing
the 'on-off' button works, and *does* turn off the IR.
I wanted to build 0.1.7 in my old environment (egcs 1.1.1) but I could not
since I didn't build the c++ cross compiler in that environment, and
legOS needs it to compile sensor.cpp.
So, something has definitely changed, not exactly for the better, between
0.1.6 and 0.1.7. I did a diff on the tm-and-ir.c files in those distributions
and the only changes were comments, so the bad change occurred in legOS
itself, not the demo.
Any thoughts?
Dwayne
In lugnet.robotics.rcx.legos, Dwayne Jacques Fontenot writes:
> I got everything to build using egcs 1.1.2, but I am having problems with
> the tm-and-ir demo (after modifying it for the new execi syntax) which
> make me think maybe egcs is not generating correct code. I used the build
> instructions at http://www.zanshinsys.com/linux-h8300/linuxgcc-h8300-HOWTO.txt
>
> I did use the --enable-target-optspace flag when building egcs. I wonder if
> that is part of the problem.
>
> I will try again a little differently to see if I get better results.
>
> Dwayne
>
> In lugnet.robotics.rcx.legos, Lou Sortman writes:
> > Luis Villa wrote:
> >
> > > I got the same message, and your reasoning about the content of the
> > > message sounds AOK by me. But then I did a make install, and that crudded
> > > out on me, with the message
> > >
> > > In file included from ../../../egcs-1.1.2/gcc/f/com.c:198:
> > > ../../../egcs-1.1.2/gcc/f/com.h:144: #error Cannot find a suitable type
> > > for FFECOM_f2cLONGINT
> >
> >
> > I did not have this problem. Sounds like Fortran. Did you 'make • LANGUAGES="c
> > c++"' or just 'make'? The latter makes the Fortran stuff, too. Remember • that
> if
> > you specify LANGUAGES when building, you should specify the same value when
> > installing.
> >
> > > > > I even tried building using newlib-1.8 to no avail.
> > >
> > > Has anyone else tried this? At least in theory, the improved efficiency
> > > would be worthwile, but I'm curious about compatibility.
> >
> > I have not.
|
|
Message has 1 Reply:
 | | Re: gcc build instructions for linux
|
| (...) I observed the same phenomenon, only I was using gcc-2.8.1. I never did find out what was wrong, but I assumed it might have something to do with the new libgcc stuff which appeared in 0.1.7. I eventually got working by installing egcs-1.1.1. (...) (26 years ago, 8-Apr-99, to lugnet.robotics.rcx.legos)
|
Message is in Reply To:
| | Re: gcc build instructions for linux
|
| I got everything to build using egcs 1.1.2, but I am having problems with the tm-and-ir demo (after modifying it for the new execi syntax) which make me think maybe egcs is not generating correct code. I used the build instructions at (URL) did use (...) (26 years ago, 7-Apr-99, to lugnet.robotics.rcx.legos)
|
11 Messages in This Thread:
 
  
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
This Message and its Replies on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
