The real problem is the Bluetooth stack. (similar to the IrDA situation) In most cases the chip manufacturer provides only the lower levels of the stack, up to the hci level in the baseband controller. To connect to a uC the rfcomm level is the simplest way. To write the higher levels of the stack is too complex for small projects. Buying a stack from an independent sw company is extremely expensive. Considering small projects, the only affordable solution if the chip manufacturer offers the full stack for free (with the chipset).

On the rfcomm level, once the connection, pairing etc. is ready, the device is transparent, very similar to a RS232 serial cable.

I found Transilica on the net as a manufacturer who advertises a high level stack for free with their chips. Unfortunately they are not ready to sell the development kit, probably because of the chip is not ready. I have some modules from Taio Yuden based on the Transilica chip, but that samples are so faulty that almost useless (reset problem, lost connections, freeze etc). I stopped to evaluate them, because Transilica refuses to give any sw support until their development kit becames available. The Transilica chip is still interesting for me because they are using a 8051 compatible core as baseband controller.

Now I decided to work with the CSR chipset www.csr.com . The development tools are available, they give good sw background, and offering the high level stack (including rfcomm) for free with their chips. Another interesting possibility is the c programmable "virtual machine". It seems to be easy to connect this thing to an embedded controller. They have an informative forum on their website too. I didn't tried this kit yet, but decided to buy one. (not cheap)