In contrast, IRAM could decide a moderate evolution of the GILDAS software. The risk is then to commit manpower to this evolution, and become decoupled from the AIPS++ suite which might become the radio astronomy package of the future.
On the long term, with only IRAM & associates (Observatoire de Grenoble, MPIfR, other French observatories), it is conceivable that the IRAM community produces and maintain a radio-astronomy package well suited to the IRAM instruments (30-m, Plateau de Bure), and also to ``Standard'' imaging with the LAMA (Large Atacama Millimeter Array, ex LSA/MMA). More fancy applications such as very large mosaics, full polarisation measurements, super-computer class images, would most likely be beyond reach.
Obviously, the IRAM community should express its view point on the future of IRAM software.