[OAI-implementers] Celestial Update & Repository Errors

[Tim Brody]

Dear all,

I've been working on improving Celestial, and as part of this have set to
harvesting the new OAI 2 registered repositories (n.b. just moved machines,
so the DNS may take a while to update):
http://celestial.eprints.org/cgi-bin/status

(red means either not harvested yet or the baseURL is wrong in the Identify
response)

Onto possible repository errors. These are all with a caveat that the
problem may be at my end - if you notice this please email me and I will
attempt to rectify as soon as possible.

1) Repositories not supporting the yyyy-mm-dd date format. I believe a
repository must support yyyy-mm-dd, even if it also supports the higher
resolution yyyy-mm-ddThh:mm:ssZ (see protocol 2.7.1)

2) Quite a few repositories appear to have Unicode errors

3) Returning an HTTP 400 error when returning OAI errors. An OAI error is a
valid HTTP response, so should be 200 (perhaps 3.1.2.2 could make this more
clear?)?

4) Poor performance leading to timeouts (my own citebase is terrible ... )

5) Incorrect baseURL in Identify response - as a harvester I use the baseURL
returned by Identify to make sure I'm harvesting from the right location

As ever, feedback, comments, flames welcome ...

All the best,
Tim Brody