"Johnny C. Lam" <jlam%NetBSD.org@localhost> writes: > A user on IRC had problems running bootstrap which was tracked > down to having MAKEFLAGS set in the shell environment to some > value that bmake doesn't understand. > > Should we modify the bootstrap script to throw an error if > MAKEFLAGS is set, and emit a suitably descriptive explanation > why? We can always unilaterally unset MAKEFLAGS within the > bootstrap script, but the user is still likely to have have > problems when trying to build packages after the bootstrap is > complete. I suppose there is no harm (post branch) to adjust the bootstrap script to be a bit more defensive (and I agree with check/error). I would rather not get into cleaning environment variables to defend against people having odd values that will lead to trouble in general, unless we take the approach that we will completely sanitize the environment and have a coherent plan for why what makes sense -- which seems like a radical step to resolve something that seems quite rare.
Attachment:
signature.asc
Description: PGP signature