We’re currently in the process of asking our Symplectic Elements current research information system (CRIS) talk to ePrints, our repository software. The idea is that academic colleagues should be able to upload full text versions of research directly to the CRIS, at which point the item gets cross-walked into ePrints automatically. This gives a one-stop-shop for academics wanting to manage their research profile and open access material, and makes administering the workflow seamless for those of us behind the scenes. Fantastic: a seamless service for academics and for administrators, and a sound basis on which to set up a new repository.
The problem seems to be actually getting this to happen. It has proved difficult to get the necessary code from Symplectic to pass to ePrints to make the cross-walk happen; then there were issues with opening up the server ports, as City’s network security people were (quite justifiably) interested in what systems were being given network access and why; and now we think it’s possible that when Symplectic uploads records to ePrints, it may overwrite material already there (fine if these are just test records as currently, not so good if genuine records we create in ePrints down the line get over-written).
None of this is to point the finger of blame, and I’ve actually not had to manage any of this directly (my colleague Andrew has had that pleasure). The point seems to me to be that systems architecture is difficult, and particularly so when you’re trying to manage a bi- or tripartite relationship (in this case Symplectic, City, and ePrints). If I were to draw a lesson from all this it would be, in a project setting, to always factor in plenty of time to allow for delays in making these relationships work, because things never run as smoothly as one would like.