[Building Sakai] Google Summer of Code is on again

[Matthew Jones]

You have to remember too, the last time that CLE leaders looked and made
presentations about an impossible list (~2008), they concluded that it was
better to start fresh than to continue with what was there. That left CLE
mostly in maintenance since then. ;)

It's possible that there *are* unknown developers out there, but some
things like this are serious months long commitments in learning,
understanding and fixing the code.

I think we also need to figure out if that cost benefit is *really* worth
it. I mean if the problem with #1 is that startup turnout is too long for
developers, then all active developers should just get computers with SSD's
and/or more memory.(16GB laptop with SSD is less than $1000 now)  That
seems like it would be cheaper than a Sakai Package Manager? (Though the
OAE Widget library is pretty cool) [1] If the problem is making it easier
for an end user to add a contrib tool, then that's a different story.

The 3 things mentioned for #2 (elasticity, failover and rolling updates)
are nice to have but rarely a problem in practice. Most clients are okay
with their systems being down for 1 minute at 5AM (updates) there is not
much cost in having systems running a few hours while waiting for sessions
to drain (elasticity) and generally when one node goes down they all go
down. It's a once in a year occurrence (if that) that a single node has a
unexpected failure. And really it's only a big problem if someone is in the
middle of an assessment or actually typing something into CKEditor and
loses that. So make CKEditor autosave (SAK-22495), turn on assessments auto
saving by default and you've basically fixed 90%+ of the failover problem.

So perhaps finding solutions that cover 90% of the impossible list are good
enough?

[1] http://oae-widgets.sakaiproject.org/

On Fri, Mar 29, 2013 at 4:12 PM, Adams, David <da1 at vt.edu> wrote:

> Ignore the Summer of Code for a moment: Zach's list is exactly the type of
> overarching goals the CLE community needs to be aiming for. Yes it will
> take a lot of work to revamp all the components that have locked us into
> the current way of doing things, but that work needs to happen unless we
> want CLE to fade away into irrelevance. Perhaps the GSOC proposals can be a
> smaller part of a larger project focused on addressing one of these needs.
> As I've argued before, if there's a clear TCC-led focused project to
> improve Sakai, I think there are resources out there who will help. Zach's
> impossible list is the first step towards making it possible. Will the the
> CLE's leaders step up for step two?
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://collab.sakaiproject.org/pipermail/sakai-dev/attachments/20130329/2bcaf0c5/attachment.html