Closed MageFroh closed 1 year ago
A few open questions that we could discuss:
Krb5ConfigDefaults.DnsCanonicalizeHostname
to a bool
value. Is that OK, or should I try to be a bit more backward compatible somehow?KerberosClient.GetServiceTicket()
, and I don't understand Kerberos well enough to do it in this PR. Is it OK to treat it the same as true
for now? Let me know if you think it should be the same as false
instead.DnsCanonicalizeHostname
is false
in Kerberos.NET, but MIT Kerberos claims it is true
. I've kept this discrepancy. But just to be sure, is it a conscious decision?Looks like the build failed at the Nuget Push stage, but I'm not entirely sure why. Any ideas about how to fix this?
Build failure is normal at the nuget stage with PRs from folks outside the org. It really should be configured not to try to push nuget packages on PRs at all, but thats a problem for another day. That said, because it got there that means the build and tests did pass. This looks good. You have added test coverage and while it is a minor breaking change to the API, I'm not terribly concerned because it's a fairly obscure config for someone to set via code. We'll bump the version to indicate. I'll merge this change and get a new package uploaded shortly.
And its live. Thanks for the PR!
Excellent, thanks for reviewing quickly!
This allows successful de-serialisation of
krb5.conf
(for example viaKrb5Config.CurrentUser()
) on Linux systems that come withfallback
as the default value.fallback
is currently treated the same astrue
.What's the problem?
Before this change, and exception would be thrown when de-serialising
/etc/krb5.conf
on a fresh RHEL 9 installation.What's the solution?
Simply allows the new
fallback
value.What issue is this related to, if any?
Fixes #329