Re: Importing Heimdal 7.8.0

Philip Paeps <[email protected]>
Newsgroups gmane.os.freebsd.architechture
Message-ID <[email protected]>
On 2024-02-05 15:17:00 (+0800), Emmanuel Vadot wrote:
> On Mon, 05 Feb 2024 14:20:34 +0800
> Philip Paeps <[email protected]> wrote:
>> On 2024-02-04 14:54:58 (+0800), Emmanuel Vadot wrote:
>>> On Sat, 3 Feb 2024 10:24:09 -0800
>>> Enji Cooper <[email protected]> wrote:
>>>>> On Feb 3, 2024, at 09:45, Piotr P. Stefaniak <[email protected]>
>>>>> wrote:
>>>>> ?On 2024-01-31 15:31:38, Dag-Erling Smørgrav wrote:
>>>>>> Minsoo Choo <[email protected]> writes:
>>>>>>> I'm currently working on importing the latest version of 
>>>>>>> Heimdal,
>>>>>>
>>>>>> Please don't.
>>>>>
>>>>> why
>>>>
>>>> Cy is importing MIT kerberos. MIT is (in many cases) the defacto
>>>> flavor of kerberos.
>>>> Cheers,
>>>
>>>  Is changing kerberos flavor in 2024 really what we want ?
>>
>> We should ship a supported / maintained flavour of Kerberos.  MIT is 
>> the
>> best option.
>>
>>> People who are using base kdc will likekly migrate to ports version 
>>> of
>>> heimdal as database isn't compatible (unless something has changed 
>>> in
>>> the past 15 years I've used kerberos).
>>
>> That's certainly true.
>>
>>> I guess that kerberos is still used a bit at some Colleges or old
>>> corporation that haven't moved from it but is it relevant for us to
>>> still include kerberos in base ?
>>
>> The kdc is only one component of Kerberos.  While using Kerberos 
>> alone
>> is certainly increasingly niche, many organisations use it in
>> combination with LDAP (e.g. Microsoft Active Directory).
>>
>> We need the Kerberos libraries in the base system for GSSAPI.  It's 
>> more
>> effort not to include the kdc and the utilities (kinit, kadmin,
>> ktutil,...) than including them.
>
>  Is there a written proposal for this switch ?

Not that I'm aware of.  Kerberos is not a particularly active area of 
the tree.  Cy has been maintaining Heimdal and has volunteered to switch 
us over to MIT.  I don't think we need any more bureaucracy than that.

> I can't seems to understand how it's useful to not include the
> utilities in base (I understand for kdc).
> If I need kerberos to login in my env I would need to pkg install
> heimdal/mit so I might as well pkg install openssh-portable && pkg
> delete FreeBSD-openssh so I have a kerberos aware ssh.

Right.  I don't think it's useful to stop including the utilities in 
base.  I don't mind not including the daemons.

We need kinit (and probably ktutil) and GSSAPI for NFS too.

I don't have particularly strong feelings about Kerberos-aware OpenSSH.  
Since we have to ship the libraries anyway, we might as well use them.

Long-term, I would advocate for "privatising" the Kerberos libraries 
(similar to what we do with sqlite3, libxml, etc) to avoid conflicting 
with 3rd party libraries.  I have no idea how much work that would be.  
I don't think I'm interested in doing the work, and I'm hesitant to 
volunteer someone else's time to do the work. :-)

> Please be aware that we're pushing pkgbase use so we will have a lot
> more flexibility to have a tool installed or not.

Sure.  And I'm all for pkgbase. :-)

Philip
lmpx.com only provides a reader for public news (NNTP) servers. It is not affiliated with the servers or forums shown here and is not responsible for the content of articles, which is written by their respective authors.