> -----Original Message-----
> From: Roland McGrath [mailto:roland@gnu.org]
> Sent: Tuesday, June 08, 1999 12:15 PM
> To: Brent Fulgham
> Cc: Debian-Hurd
> Subject: Re: strace/ltrace
>
>
> > Does anyone know what would be involved in implementing the
> > Linux strace/ltrace features on the Hurd? I would find
> > these to be tremendously helpful in tracking down problems,
> > but I don't know what the level of effort involved would
> > be. Any comments?
>
> ltrace and strace are quite different things.
>
> I had not before known about ltrace (though naturally it was
> already installed on my Linux system!), and just now looked
> briefly over it. ltrace should not be hard to port to the Hurd.
> It mostly uses the ptrace interface, which also exists on the
> Hurd (though of course it hasn't been tested). It would also
> not be hard to port it to use the native Mach and Hurd interfaces
> for the things it does (use the code from gdb as a model).
>
That was my initial thinking as well. That's why I think I may
be able to get GNU/Nana (or at least a useful subset) to work, too.
I am less optimistic about my abilities to get strace done in
a short period of time (although it would definitely help me
learn the hurd internals). If I can get the ltrace and nana in
working order, I will try strace next (unless others have
made significant progress).
Would revising ltrace to use hurd-ish techniques provide
any advantages over the ptrace interface?
-Brent