>> try to omit -pipe... ;)
>
> Really? How curious... trying. That would be weird though, it's not=20
> meant to have a
> negative effect on systems that do piping properly (if NetBSD didn't=20=

> this would have
> come up already :P), and to stop lex from doing its thing?
>
> I had similar problems, I recall now, in a Gentoo rig (yuk) where -j>1=20=

> was used, and
> Makefiles lacking .WAIT did things in the wrong order and tried to use=20=

> a header
> before it was generated. However, here it defaults to -j1 (right?) and=20=

> forcing it
> didn't make any difference.
>
> Trying with -O -march=3Di686 and nothing more, and after the set of=20
> hacks suggested by
> Adam Ciarcinski.
>
> Thanks for the quick replies by the way, I was expecting to have to=20
> wait until
> tomorrow :)

as can be seen from dmitri's 16:06:00 +0100 (CET) email, omitting -pipe=20=

did the job.

furthermore, yes, on my MP machine (unfortunately an i386 yet =3D8-) -j =
n=20
(with n=B32) works quite seldom...

--=20
:x!

timo