[distcc] Re: safeguard and stuff

[Wayne Davison]

On Wed, Apr 02, 2003 at 03:24:38PM +1000, Martin Pool wrote:
> I can't immediately reproduce this but I haven't tried very hard yet.
> Perhaps it's something to do with your path or masquerade setup?

How silly of me, that's exactly what was going on.

> Do you mind explaining what the meaning of safeguards>1 is, and how
> they're meant to interact with dcc_support_masquerade()?

The idea is that we can allow safeguard to be 2 (indicating that either
distcc or distccd called distcc unexpectedly) if we're making progress
somehow, i.e. if we trimmed the PATH and thus believe that the next time
we call gcc we won't just encounter ourself again.  This does allow a
slight inefficiency in the setup to go uncommented, but the new PATH-
tweaking patch for distccd should eliminate this.  If you make the non-
masquerade mode tweak the PATH in the same way as the distccd change,
you could probably change this back to always complaining about any
type of recursion (instead of allowing one inefficient callback).

> Also please stick to returning EXIT_* values in anything you should
> happen to write in future.

Sure.  I'll update my patch with this in mind.

..wayne..