[1012] in Kerberos_V5_Development
Re: Getting educated
daemon@ATHENA.MIT.EDU (Sam Hartman)
Tue Feb 20 16:18:13 1996
To: Richard Lewis Haggard <haggard@world.std.com>
Cc: (Sam Hartman), krbdev@MIT.EDU, pbh@MIT.EDU
From: hartmans@MIT.EDU (Sam Hartman)
Date: 20 Feb 1996 16:17:35 -0500
In-Reply-To: Richard Lewis Haggard's message of Tue, 20 Feb 96 13:53:31 EST
>>>>> "Richard" == Richard Lewis Haggard <haggard@world.std.com> writes:
>>
Richard> Sorry to take so long in getting back to you.
Richard> I'm working as a consultant at OpenVision porting krb5b5
Richard> into the windows NT environment. Along with the libkrb5
Richard> port we also want to have the standard clients up and
Richard> running. Hence, I have no choice. I take that code as it
Richard> comes and work with it. In the process of doing so I am
Richard> running into more memory leaks and odd things than I know
Richard> what to do with. Our in house expertise is no longer
Richard> available and so I'm having to become the in house
Richard> expert. This is something that I'm not at all adverse to
Richard> except it seems that the only way one acquires the
Richard> necessary knowledge is by being a part of MIT's kerberos
Richard> team to begin with and working on it for a couple of
Richard> years. I haven't a couple of years and, besides, I'm
Richard> already working on getting rid of any excess funds by
Richard> throwing my educational dollars away up the street at the
Richard> big H school of overblown reputations.
I would strongly advise against using beta5 as a base for any
new development or porting. Our current code is significantly more
stable, although it lacks a working kadmind.
Depending on how far into the project you are, it might be
worth your time to obtain a pre-release of our current clients and
library, as they have about a year's worth of bug fixes.
Also, I believe people in our DOS/Windows development group
have tried to get Kerberos5 working on NT, and some of the necessary
changes have made their way back into our tree.
As far as learning the API, I basically ended up listening to
some impromptu lectures on how Kerberos worked and then reading the
API documentation through. It's lacking a good description of the
flow of events and a description of what is needed when.
Richard> And now for something completely different- I've got all
Richard> of this lovely krb5b5 code that I'm making alterations
Richard> to. Is there an established mechanism for passing the bug
Richard> fixes back? Who would I talk to about such things?
There is a mailing list krb5-bugs@mit.edu where bugs and
patches go. You can be added to this list if you wish so you will
receive other patches.
If you are interested in looking at back-archives of this
list, they are available in a discuss meeting at
menelaus.mit.edu:/usr/spool/discuss/krb5-bugs.
Richard> Richard Lewis Haggard
