It's common, especially amongst the Austrian types on this forum, to use inflation in it's original (monetary only) meaning, while others (typically the Keynesians and monetarists) use it in it's more modern price inflation meaning. If we're using the modern meanings of inflation and deflation, then Bitcoin is very strongly deflationary at the moment. If we use the original terms, then it's strongly inflationary, but demand is increasing even faster than supply.
It happens that any non-monetary "definiton" of inflation fails to be a definition and practically works as either a red herring or strawman in any discussion of economics. Should somebody finally be able to construct a non-monetary definition that actually works as such it may be taken into account (at some point, by someone). As it is, the keynesian pseudo-inflation is pretty much only interesting to politicos.
I agree completely. I count myself amongst those "Austrian types."
Eventually, Bitcoin will switch to being properly deflationary, but given that a) it will be a very gradual slope, or at the very least very long plateaus between deflationary events (lost or destroyed private keys) and b) lost and hoarded coins are functionally identical, and unless a private key is known to be lost, it's unwise to treat it as actually gone, I don't see it being a problem. When/if it does become a problem, Bitcoin2 is a simple code fork away.