pitman at stat.Berkeley.EDU
Fri Feb 3 16:30:53 UTC 2012
(previous subject was "OpenBiblio call tomorrow")
> On 1 February 2012 15:38, Jim Pitman <pitman at stat.berkeley.edu> wrote:
> > My big concern is that despite recent advances on other fronts we have two target milestones from some months ago
> > which have not been achieved:
> > a) installation of at least one public facing bibserver on a machine not controlled by cottagelabs/OKF
> These aren't a priority without some obvious committed person who wants to said installation and deployment.
We are bulding cars and we dont think its a priority to find drivers?
> Several devs have deployed it without a problem.
Sure, but this is just devs making work for themselves. The goal of the project is to create bibliographic
software that is actually used by bibliographic data curators. So we have to bridge from the devs to
the curators, and this has not happened yet.
> In my opinion, providing a good hosted option (be that bibsoup itself or hosted instances) are much more important than this.
Strategically I think that is right, but only because of the big hump we have found in finding hosts to deploy
and configure BibServers for their purposes. Longer or even medium term, we should still be looking for other maintainers
> > b) customization at least one bibserver for specific editorial purposes, to suit the needs of some biblio curator.
> Again we need some clear lead and some clear commitment from a third party to do this.
I am the lead and third party for Probability Web deployment, which Mark and I have agreed to proceed with.
But we need to be looking aggressively for other leads and commitments.
> > I think these should be at the top of current priorities.
> > It seems customization of a bibserver should best be done by cottagelabs/OKF, with a longterm SaS model, as the hurdle for
> > a) seems higher than anticipated. We have to learn what are the obstacles to a) and how to get around them.
> a) isn't that hard but it depends on having a sysadmin willing to deploy ES and python.
We havent been able to do a), so operationally it has been hard.
The difficulty is more social than technical. The sysadmin must be authorized by their employer to deploy a BibServer on
behalf of their organization. This requires administrative approval, and an editor/curator to oversee the BibServer. So you need 3 components:
1) sys admin capability and willingness
2) administrative approval from organization
3) editorial capability and willingness
No two of these suffice, and that makes a problem. As we gain experience with deployment, we can hopefully
find ways to bring these components together. And in the meantime, I am fine with the SaS route.
Professor of Statistics and Mathematics
University of California
367 Evans Hall # 3860
Berkeley, CA 94720-3860
ph: 510-642-9970 fax: 510-642-7892
e-mail: pitman at stat.berkeley.edu
More information about the openbiblio-dev