need feedback: immediate move of Atlassian Confluence Shibboleth Authenticator to new repo

[Gary Weaver]

Peter,

Thanks for the feedback!

I understand that many aren't avid git users. I was not at first years ago
after having used svn and then hg for a while, and it still depends on the
project(s) of course, but if you haven't tried spending a month or so with
git and GitHub on a personal project, I recommend trying it out. It isn't
perfect, but I've come to like it more than svn or hg, at least.

I also understand first-hand that users don't fork and contribute just
because it is on GitHub, so I'm not under that illusion either. And I don't
want to choose git if the community is going to be less likely to
contribute to it.

The main reason I would choose GitHub as the provider is that between it,
bitbucket, sourceforge, Google code, and java.net, GitHub seems to be the
biggest, most stable, open source project free hosting provider that I'm
aware of. I've had projects hosted on GitHub and SF and that's mostly
because I've not been a fan of the other interfaces- although bitbucket is
a close second to GitHub).

Like you I'm in favor of longevity of this project with as few forced moves
of the repo as possible, so I would choose GitHub as the provider if
neither you nor others that would contribute to the project would have
strong feelings against it.

Sorry was not aware of the Internet2/Shibboleth split, and that doesn't
affect my inclination to assist in my spare time.

I also agree that the authenticator is not Shibboleth-specific, although
the contributions from Bruc Liong some years ago were geared toward
auto-creation/management of groups (role) based on headers that were
populated by mace attributes, but again even that is not necessarily
Shib-specific. If anything, it has just been that the community afaik that
uses and supports the authenticator tends to be primarily the Shibboleth
user community. Chad's original name of the authenticator was
remoteUserAuth (although with Craig Solinski's contribution, it isn't even
tied to REMOTE_USER header anymore), and it could just be called "http
header authenticator" and I could just provide the shibboleth config info
to try to keep from losing any future Shibboleth community of users that
could better take advantage of it with that info?

So, given all of that, what if I just created a group in GitHub:

https://github.com/confluence-http-header-authenticator/

and then call the new project "Confluence HTTP Header Authenticator":

https://github.com/confluence-http-header-authenticator/confluence-http-header-authenticator/

Then anyone who would like to be a member of this group (and not just
fork/pull-request) just contact me and I'll add you because I don't want to
be the sole person with rights to the repo, even if I'm the only person
interested in maintaining it, which hopefully I'm not, because my time is
really limited.

Let me know your further thoughts, and thanks for the info on the list
subscription/configuration.

Thanks!
Gary

On Tue, Jan 15, 2013 at 10:24 AM, Peter Schober
<peter.schober at univie.ac.at>wrote:

> * Gary Weaver <garysweaver at gmail.com> [2013-01-15 15:37]:
> > So, first: does anyone have a problem with hosting this project at
> GitHub?
>
> Personally I'm not an avid user of git and I also don't share some
> people's conviction that "put it on github and they [i.e., developers]
> will come" but I'm certainly not against such a move.
> Any repo you control (for varying degrees of "control") is preferable
> to being chased out of some infrastructure every other year or so.
>
> > Does anyone know who owns the internet2 group at GitHub or who has a
> > similar group (shibboleth, etc.) that might make a good place for the
> > Confluence Shibboleth Authenticator project?
>
> I don't, but note that this choice for a new home seems pretty
> arbitrary now that Shibboleth is not officially an Internet2 project
> anymore. Cf. http://shibboleth.net/consortium/
>
> > I would even say that maybe the Internet2 project could host it, but
> > I like the forking/etc. utility provided by GitHub and think it is a
> > much more easy and collaborative way to handle open-source projects
> > than other alternatives based on my experience so far.
>
> Same for the SVN repositories as used by Shibboleth (i.e., they're not
> being hosted on Internet2 infrastructure anymore).
>
> > Anyway, if I don't hear back, I may end up just create a new group
> > in GitHub like "atlassian-shibboleth" or similar and put it there.
>
> Is that code really specific to Shibboleth? If not (since all it sees
> are HTTP request headers -- making it also possible to use request
> attributes would be a desired improvement, btw) I would advise against
> using "shibboleth" in the name.
> Speaking of names I'd also be wary of the use of "atlassian" as part
> of the name -- who knows what else they might change their mind about
> in the future?
>
> > This is your project and I've just been helping as I can and want to make
> > sure that it doesn't disappear in a few weeks.
> >
> > Let me know if you have any feedback, and note that I'm no longer working
> > for Duke Univ., that I'm just doing this in my spare time now, and I only
> > become a member of this list periodically to send a shoutout for things
> > like this, so please contact me directly at my gmail address in the
> future.
>
> Thanks for your continuing efforts.
>
> Regarding periodically (un)subscribing lists note that you can disable
> mail delivery from the lists' web interface if you don't intend to
> follow, cf. http://shibboleth.net/mailman/options/dev ("Mail delivery"
> -> "disabled", with "Set globally" ticked) while still being able to
> post (as a member/subscriber of the lists).
> -peter
> --
> To unsubscribe from this list send an email to
> users-unsubscribe at shibboleth.net
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://shibboleth.net/pipermail/users/attachments/20130115/7d8a63fb/attachment-0001.html