[OAI-general] System Architecture
Fri, 14 Feb 2003 15:45:24 -0500
DSpace uses our OAICat implementation to enable OAI access to the DSpace
database. There are a couple of possibilities here.
The first is that the data as it exists in the DSpace database contains bad
characters. If that's the case, there's nothing OAICat can (or at least
should) do to correct the situation. Basically it's garbage in, garbage out
at that point.
The second possibility is related to the fact that OAICat is a framework
that can be customized to work with any database that supports a Java API.
This is done by implementing some abstract Java interfaces to the search
engine of one's choice. This was the case with DSpace, where Robert Tansley
wrote some code to interface with JDBC. It's possible that something is
amiss within this custom code.
I don't believe that the problem is in the OAICat framework itself because I
have a couple of OAICat repositories that produce proper Unicode results.
Unicode is such a pain.
> -----Original Message-----
> From: Mark Jordan [mailto:firstname.lastname@example.org]
> Sent: Friday, February 14, 2003 3:17 PM
> To: Young,Jeff
> Cc: email@example.com
> Subject: Re: [OAI-general] System Architecture
> On Feb. 6, Tim Brody claimed on this list that DSpace wasn't
> OAI compliant based on harvesting attempts he'd made (see
> y/000239.html), suggesting that DSpace allowed
> XML-illegal characters where the harvesters expected legal ones.
> So, I guess my naive question is... Is DSpace OAI compliant or not?
> On Fri, Feb 14, 2003 at 09:38:21AM -0500, Young,Jeff wrote:
> > Amy,
> > DSpace is a wonderful product, but OAI-compliance is a
> very small part
> > of what DSpace is about. If you merely want to add OAI
> capability to a
> > system that already exists, DSpace would be overkill and would
> > probably make your job much harder. Because DSpace is an fairly
> > elaborate system, using it as a model for your own architecture
> > probably wouldn't be appropriate either.
> > OAI-compliance can most easily be achieved by
> downloading one of the
> > OAI implementations found on
> > http://www.openarchives.org/tools/tools.html. This
> page is a little
> > misleading, though, because many of the tools listed are
> much more
> > than mere implementations of the OAI protocol (e.g. DSpace).
> > I'm only familiar with two of the implementations that
> are available.
> > If you're most comfortable working in Perl, the VTOAI
> OAI-PMH2 PERL
> > implementation.seems to be
> > popular
> > tml.) If you're most comfortable using Java, I would suggest
> > OAICat (http://www.oclc.org/research/software/oai/cat.shtm)
> > which runs under a J2EE-compliant servlet engine (such
> as Apache's
> > Tomcat). I am the author of the OAICat and would be glad
> to help you
> > get it set up for your needs. BTW, DSpace uses the OAICat
> > implementation to provide its OAI repository functionality.
> > OAICat can be customized to work with any database engine that
> > supports a Java API. Implementations are available for
> JDBC as well as
> > a simple file-based system. Other databases can be
> accommodated by
> > implementing a set of abstract Java classes.
> Implementations are also
> > available for Newton and the open-source Pears database
> engines which
> > are used in OCLC's SiteSearch product.
> > BTW, if you have an unusually large repository (or even
> if you don't)
> > you may even want to consider using the open-source Pears
> > database
> (http://www.oclc.org/research/software/pears/). I use it
> > for our XTCat repository (http://alcme.oclc.org/xtcat/) which
> > contains over 4.2 million records.
> > Hope this helps.
> > Jeff
> > -----Original Message-----
> > From: Hatfield, Amy J [mailto:firstname.lastname@example.org]
> > Sent: Friday, February 14, 2003 8:50 AM
> > To: email@example.com
> > Subject: [OAI-general] System Architecture
> > Hello -
> > I am currently exploring our institution's ability to create an
> > institutional repository that is OAI compliant. I feel
> I have a fair
> > understanding of the standards, functionality, etc.
> involved - and am
> > leaning toward DSpace as a model - but there is one aspect that I
> > don't see discussed very much... system architecture!
> > I would be grateful if folks would share with me some
> > configurations they have developed. I am most interested in the
> > storage aspect. I have looked at jukebox technology as a storage
> > backend, but am not sure about the retrieval aspect. We
> also have a
> > super computer with lots of storage capacity - but it is
> not designed
> > to handle small files, but rather large datasets. Any
> information you
> > would like to share will be most appreciated.
> > Feel free to respond directly to me if you would rather
> not post to
> > the list.
> > Kind regards,
> > Amy
> > Amy Jo Hatfield, Systems Librarian
> > Ruth Lilly Medical Library
> > Educational Technology
> > (317)278-8402
> > 975 West Walnut Street
> > IB-100 Room 102A
> > Indianapolis, IN 46202-5121
> > References
> > 1. http://www.openarchives.org/tools/tools.html
> > 2. http://www.dlib.vt.edu/projects/OAI/software/vtoai/vtoai.html
> > 3. http://www.oclc.org/research/software/oai/cat.shtm
> > 4. http://www.oclc.org/research/software/pears/
> > 5. http://alcme.oclc.org/xtcat/
> Mark Jordan
> Acting Coordinator of Library Systems
> W.A.C. Bennett Library, Simon Fraser University
> Burnaby, British Columbia, V5A 1S6, Canada
> Phone (604) 291 5753 / Fax (604) 291 3023
> firstname.lastname@example.org / http://www.sfu.ca/~mjordan/