What problems do we need to address with a repository?

[Ben O'Steen]

Scott nailed it on the head earlier:

"perhaps the question
should be more reflective... why do we care, and what do we really
want to achieve? If lightweight repositories are the proposed
solution, then what is the problem they are supposed to address? "

Putting myself in the user's shoes, I would want somewhere to put my
files, so I don't have to bother about things like backups, usbkeys,
or making sure that people I want to work with are on my institution's
network so that they can see the files I want them to see. I want a
store that would help me keep track of the research I am doing, from
building citation or bookmark lists to roping together the two google
docs, slides on slideshare and images on flickr that I am using to
draft a proposal. It's a side effect to me that a system might backup
all these things on demand too, and it's definitely a side effect if
it makes it easier for my institution to 'claim' my work and make it
visible through their repository, if I wish it.

Changing hats now, to that of a repo manager, I would want a system
that made it as easy as possible for people to register their work for
open access or for archiving. With a research services hat on as well,
it would be fantastic to join the funded projects to the project's
disseminations. I think that other hats could be imagined here as
well ;)

So, what is the problem that the lightweight stores aim to help with?
It's the current problem of getting content. A lightweight store would
provide useful things to the academic/researcher/etc which is crucial
in it being adopted, so this should be the main focus. Current stores
lack certain features which would make the content they hold much more
useful to other systems, and this is something else to be addressed. A
lightweight store is also not an institutional view either, so there
is no pressure for it to capture and indicate the canonical, super-
duper, final version of works only.

But it should have enough features for other systems to hang off of it
- for example, noone would suggest that the mysql commandline is a
useful UI to every user as is, but you can understand the sheer
variety of systems that can and have been built using the simple
commands of SELECT, UPDATE, DELETE and CREATE (i'll go into what I
think are useful baseline features for a lightweight store in the next
post, one about REST-ORE, restore, geddit?)

So, while I have got an unintential bias to solving some problems for
myself in the job that I do, I am more interested in discussing, and
perhaps building, what I see as truely useful web resource storage.
Whether this is an FTP box, with an API to register for changed/
updated files, or a WebDAV with a smart, file-aware, search engine
builtin or a REST-orientated dropbox for files, with a search engine,
using atom or oai-ore to bind groups of resources together, atom for
ordered, ore for unordered, with a messaging API on top, to push
events out or to be polled for events that have occurred. (The last
one is a synopsis of REST-ORE ;))

Ben O'Steen