Re: Adding makemandb as build tool for man.db

[David Brownlee <abs%netbsd.org@localhost>]

On Wed, 31 Oct 2018 at 20:55, David Riley <david%the-rileys.net@localhost> wrote:
>
> On Oct 31, 2018, at 3:07 PM, Martin Husemann <martin%duskware.de@localhost> wrote:
> >
> > This would include the makemandb hint (which actually is not *that* painfull
> > on most small / slow machines I have), and also point at something like
> > this to be added to /etc/ssh/sshd_config:
> >
> > HostKeyAlgorithms       -ecdsa-sha2-nistp256-cert-v01%openssh.com@localhost,-ecdsa-sha2-nistp384-cert-v01%openssh.com@localhost,-ecdsa-sha2-nistp521-cert-v01%openssh.com@localhost,-ssh-ed25519-cert-v01%openssh.com@localhost,-ssh-ed25519-cert-v01,-ecdsa-sha2-nistp256,-ecdsa-sha2-nistp384,-ecdsa-sha2-nistp521,-ssh-ed25519
> >
> > KexAlgorithms           -ecdh-sha2-nistp256,-ecdh-sha2-nistp384,-ecdh-sha2-nistp521,curve25519-sha256,-curve25519-sha256%libssh.org@localhost
>
> Are the X25519 options really that slow? I'd expect them to be faster than medium-sized RSA keys, in general.  The NIST curves are less speed-friendly.
>
> Or is it the SHA2 part that does it?

I wonder if it also varies somewhat by cpu - there might be some cases
where a slower cpu will perform relatively better compared to a
nominally faster cpu on a certain algorithm based on the cost of
different operations.

I think the ssh optimisations sound like an excellent idea - could I
suggest splitting it out onto a thread with a different subject (and
potentially different set of list
cc's) to get more attention and then push the result back into a
tech-install thread? :)

David