Re: After enabling IPv6 curl github.com fails, succeeds with IPv4

[Mayuresh <mayuresh%warunjikar.in@localhost>]

On Tue, Jul 21, 2026 at 09:42:06PM +0200, Martin Husemann wrote:
> I guess this is worth a PR.
> 
> The resolver behaviour is suprising at least and there should be a way
> to turn it off (plus as Christof pointed out, it may do a second
> totally useless identical query).

As pointed out by Steve and beaker, resolvconf.conf has a way to configure
resolv.conf contents.

While the behavior of resolver may come as a surprise to many (e.g. it
manifested in `git pull' failing, then the obvious attention went to
whether openssl is functioning etc. before realizing the issue), it is
still a documented and controllable behavior.

Just wonder whether the default should be conservative i.e. do not retry
by appending, unless a directive says so. Also wonder, how other systems
(e.g. Linux, FreeBSD) behave in similar situation and is there any
standard like POSIX for the resolver behavior.

But retrying the identical lookup is a problem anyway (mentioned in
Christof's mail). I'll file a PR for this. Request views on the above
point, that will help decide the scope of the PR.


-- 
Mayuresh