Just thought I'd throw in a couple of points...
We are currently carrying out a survey of over 3000 PhD students, for which we are using the survey package under our main dotLRN instance.
The point here is that these students are not existing users, nor will we expect them to necessarily use our dotLRN system once the survey has completed (we use the community aspects, not the course management features). When the survey is finished, we will have all these users in the system with no way to remove them permanently. Not a huge issue, but some bits of the system might get sluggish with a large enough number of users.
I guess the process of getting students created and registered with the appropriate class (using existing data) is what most people/universities moving to dotLRN would be interested in doing?
If so, I noticed that bulk registration of users in dotLRN is quite slow, so even if you could pick out the relevant details (user, course, class) to use, moving a large number of students into dotLRN could be lengthy!
Admittedly we are using an old version of dotLRN, and I would be interested in how the latest version compares. It takes about 20 seconds to create the user, register them with dotLRN, set the browse permissions and add them to a community/group.
Overall, I think the issue with keeping users may be more about performance than about storage space, plus the point about contributions is a valid one. Perhaps there could be some way to delete users who don't have any content associated with them (like ad_remove_user for ACS)?
<Raad>
