> >Has anyone done a proper job of fixing suidperl for NetBSD current?
> >
> >If not, I'll do it.
> 
> it may not be worth it - perl5 is almost beta, and from the
> looks of it, it's not going to be beta for very long.  there
> won't be another perl4 release, so unless your fixed version
> goes in to the othersrc tree, it may not be worth it.

I originally built suidperl, but the last time I rebuilt it I just
left it out.

Curious thought...could NetBSD use the fdesc fs to have more secure
setuid scripts like SVR4 does?

Ala, kernel runs setuid script.  If the /dev/fd is available it puts
the opens script on fd 3 (or whatever is available), and then passes in
/dev/fd/3 as the argument to the script.  If /dev/fd isn't available it
does it the old way.

> on another note: whats the proper way to deal with perl's
> use of semctl() that conflicts with the netbsd header files?
> (i don't have the error message handy, something about the
> last argument perl passes not being a union semun or
> soemthing).

I fixed this.  I don't remember what I did exactly (I just moved my modem
to a different com dev and I can't login back to home), but I think
it involved just adding the proper typecast (if you look at what its doing
it's correct, the problem is older versus newer types/protos and Classic 
versus ANSI C).

I also ended up just manually editing all the AF_LOCAL's to AF_PLOCAL or
something like that.

-Mark

------------------------------------------------------------------------------