[dns-operations] a note on fetching the root zone using "dig"

Mukund Sivaraman muks at mukund.org
Sat Nov 3 05:22:51 UTC 2018

On Fri, Nov 02, 2018 at 04:36:54PM +0530, Mukund Sivaraman wrote:
> > so back in BIND9 V9.10 when idn support was added, the default for dig
> > should have been "backward compatibility", that is, no IDN output unless the
> > user asks for it AND the libraries supported it.
> Distros like Fedora (and I guess RHEL/CentOS too) have defaulted to IDN
> being on in BIND RPM builds for years now (even before this IDN code
> refactoring and addition of +noidnout). Previously one had to set
> environment variables to turn it off.
> I agree it should be off by default, as some things like loading a dig
> +onesoa axfr output would have issues even when there are no conversion
> errors.

Sitting at this IETF Hackathon, I removed IDN support from Loop. It was
implemented in a hackish way as a custom output filter anyway, that
needed per-thread state and it felt great to remove all of that. I don't
think too many people will miss it even if it were removed from BIND. I
guess BIND will hold on to it as it tries to be more of a reference

 216 files changed, 6 insertions(+), 81197 deletions(-)

 (that includes an idnkit from contrib/)

I guess a hackathon can also be used to remove things. :)


More information about the dns-operations mailing list