I've done it twice, but only when a student specifically approaches me and we agree ahead of time on the work effort. I would suggest if you do though, you reserve the right to interview ahead of the assignment due to the usual policies regarding system access in informatics.
I usually have them do a bunch of background training (send them to the Pioneer Data Systems Greg Kreis class for M and the Intersystems training for Cache for Cache.dat and Cache class instantiation) before the rotation and then when they arrive, send them with a longitudinal "fix the clinical reminders or order sets" or "here's some Forum (alpha and beta) IOC code that needs regression testing" or "I have this data that I need to get into HL7 format, can you write and test the JSON or XML to format the message from VistA".
This has both the intention of warning off anyone who doesn't want to work hard in the rotation, but providing an invitation to those who want the hardcore introduction training for system maintenance and development. If they are not federal bound, I have them get the Intersystems background and train them more on the protocol transmission systems to develop messages to go to CMS or the other regulatory matters. If they are federal bound, they have the choice of learning VA DHCP (RPMS and BEMR side training included if that's where they are going), CHCS II/CHDR if they are going DoD, or CMSNet if they are going to Beltway HHS.
And those of you who are VA, this is @ access in VistA both on the test Canandaigua and the Clin1 IOC side or complete National access to AITC's systems for that training. No better way than to jump into actual testing and debugging work. Much more intensive than the ADPAC standard day, but I find most ADPACs to not care enough about data corruption from bad practices such that I always aim for the low-level training to get mine to understand the implications.