[TOS] Infrastructure Team: Can haz?

[Karsten Wade]

Chris' email presents enough of the cross-discussion so far that I'm
going to find a way to say everything I have about this thread in this
one reply.

On Wed, Jan 12, 2011 at 09:06:43AM -0500, Chris Tyler wrote:
> 
> Karsten had talked about heading this up and doing something with
> OSUOSL, but I haven't seen anything further on this for a while.

Yep, and in hindsight it was a bad idea for me to accept the role
overall.  Aside from the lack-of-time constraints, I also don't have a
single vision on how to proceed.

The one vision I did have, and enacted, is a forum for the "Practical
OSS Exploration" textbook, which we setup in a way that it can be a
subdomain of teachingopensource.org.  It's not yet enabled for use,
but only needs a wee bit more work.

This leads me to ...

> There are three separate questions that need to be addressed:
> 1. hosting - where is the TOS.o infrastructure going to be hosted?
> 2. team - who's going to be involved, and in what ways?
> 3. services - what is the TOS.o infrastructure going to provide?

4. What are services that TOS overall as a project itself need
compared to services needed by TOS initiatives such as POSSE?

The forum for the textbook is a good example.  It requires a sysadmin,
or an OSUOSL-type of situation, or a lightweight team working a
Dreamhost-one-click-app-type solution.

TOS wiki, planet, and mailing lists are other examples.

That we sometimes use a POSSE-specific planning list, within the TOS
domain.

So when I read Mel/Sebastian's request, I saw a mix in there, but
mainly I think they are looking at resources that *aren't intended to
be in place of a FOSS project*.

> On (1), I'm approaching Seneca about hosting TOS.o there, in the absence
> of forward movement elsewhere.

I'm nervous about handing off to an unknown IT entity, in terms of
responsiveness.  If we want raw hosting, I think we have already
decided on this mailing list last year to move raw hosting to OSUOSL
simply because we could define and share contact roles more easily.

To put an example on it, if we use Seneca hosting, can you designate
external contacts who can authorize everything?  Can those external
contacts authorize other external contacts?  Can we easily setup
sub-domains?  Etc.

> On (2), I've had several students ask about volunteer
> sysadmin/netadmin opportunities involving open source.  Double win:
> experience for some students and labour for the team :-) They will
> need some leadership, and I'm willing to put some personal cycles
> into that.

I think growing the "train-up as a sysadmin by supporting
tos.org"-program is a great idea, and one of the reasons I want to see
tos.org hosted somewhere that we can be flexible with access and such.
Maybe we use full-featured OSUOSL for now, and in X months we get a
limited sudo access for student sysadmins to begin with?  (That's
similar to what Fedora Infrastructure does, iirc.)

> On (3), I really (really, really) don't want to see TOS.o create any
> sort of sandbox functionality that duplicates what the open source
> communities are providing nor insulates students and faculty from
> involvement in real projects.

So we've merged two different discussions in to this conversation:

1. What infra does a POSSE need every time so it can be successful?
2. What infra does TOS want to provide its community?

On the second, we provide a wiki, this mailing list, a Planet, etc.
We could or could not provide other special services at-a-click for
TOS community members.

On the first, I think this discussion has gone in to a logical or
assumption problem of some sort.

The points about "making your own project is a problem" doesn't
pertain to POSSE because the nature of POSSE is to get involved in an
existing project during that week.  It is simply impossible to fall
into and create the impression that instructees should be going off in
to a sandbox instead of using a real, breathing project ...

... but sometimes you want a quick service to show how it's done.
Literally a setup and throwaway to show what a technology looks like.
Especially technologies that aren't represented by the FLOSS project
that is part of the POSSE, so you can do a comparison/contrast.  The
kind of thing that makes you tempted to setup your laptop as a server
for the classrom just to show them how Foo is done ...

The list from Mel is a bit mixed, too, in that it could all be handled
by an all-in-one infrastructure team as she specifies, but it doesn't
have to be.  There ARE items that we need to figure out how to provide
and not rely upon our partner FLOSS project to provide for us:

* TOS wiki uptime.
* POSSE live/remix.
* Etherpads for the POSSE week.
* Newcooltechnology Thisweek 1.6.2.

... and some that no one provides that I know of, but are useful for a
class, etc. (I ran in to one of these recently myself, helping a high
school journalism class & trying to figure out how to aggregate their
wordpress.com blogs without hand-building a Planet.)

* Planet for a class.
* Etherpad for a class.
* Etc.

Also, we're all sitting here in our armchairs with tons of our own
classroom experience, but let's give Mel additional credit as the most
POSSE teaching of us out there.  If she says so, I'm willing to agree
we have a problem in need of solution, even though I'm not sure yet
what that solution should be.  Therefore, I don't agree with the idea
that there isn't a problem, the FLOSS projects have all we ever need,
and we don't want to hold professors' hands forever.  I trust that
POSSE is an unusual teaching/learning experience as compared to all
others that we collectively know of, and it may require something
special from that.

> It is *far* too easy to go off in a corner, play on a toy project (that
> no one will ever care about) using GIT and place it under an open source
> license and pretend you're doing open source; I would far rather see
> students jump in to an actual community that's doing something real.
> There are plenty of opportunities at every level of involvement and
> skill to absorb thousands of students, especially if faculty take the
> time to get involved in the community first.

I think this is baked in to the POSSE curriculum and I don't see us as
being in danger of breaking this rule.  However, if it helps, perhaps
we could have some specific TOS guidelines to our future selves that
reinforce this idea, so we don't in the future start providing
gitorious-in-a-box for classrooms where they should be working with
SomeReal Project.

> Having said that, I'm not convinced about the value of things like git
> repositories or trac instances, since those things are provided by the
> communities (if you're hacking LibreOffice, use their infra; if you're
> hacking Firefox, use their infra; ...

I want to show a professor how to clone a git repo from LibreOffice
and make it available on another host so that others can clone from my
branch and check out the cool stuff I'm doing without it having to be
merged back to LibreOffice.  So _today_ I can do that with
gitorious.org, but last year-or-so I couldn't.  I would have been
messing with httpd and stuff.

Say I want to show how easy it is to setup a yum repository for your
packages before they get submitted to Fedora, today I can do that
using fedorapeople.org but only if I can install the Fedora packaging
and yum devel tools locally.  What if the professor brought a Windows
or OSX machine?  I sure would like to boot to a Fedora spin or POSSE
remix ...

What I'm saying is that I can think of a huge number of ways to remix
FLOSS technology that might benefit from seeing how something is
actually done rather than looking for an existing project that hosts
it as a service or provides it as part of their FLOSS infrastructure
for contributors.  I wouldn't summarily close the door on the idea of
POSSE instructors being able to spin-up needed services because of the
risk that it might make people think we are recommending they just go
in a corner and do their own thing.

> if you're creating your own open
> source project, **YOU'RE DOING IT WRONG**, and there are lots of places
> that will give you infra, such as github and sourceforge and
> fedorahosted) -- though I waver on that and can see that there might be
> some useful places for those types of services. I can definitely see the
> value in providing things such as Planets.

I think there will always be a new FLOSS technology that needs a quick
example or dirty hosting for the week that is not provided somewhere
else.  How to get that is still a bit murky, although I do see the
Dreamhost-style as being close (I use this for my personal projects,
for example, and it works well for that.)

- Karsten
-- 
name:  Karsten 'quaid' Wade, Sr. Community Gardener
team:                Red Hat Community Architecture 
uri:               http://TheOpenSourceWay.org/wiki
gpg:                                       AD0E0C41
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
URL: <http://teachingopensource.org/pipermail/tos/attachments/20110112/477ea41c/attachment.asc>