If by fusion you mean hot fusion, the kind done in Tokamaks, then yes, we should forget it. This is a terrible waste of money--- the designs are not going to produce a power plant in 2050 or 2080 or 2100, or 3000, or anytime. It was a good idea, it just didn't pan out, and the scientists involved are constantly afraid of losing their funding, something which should have happened a long time ago. The major issue is that you don't get factors of 100 in efficiency from scaling up, so if the current billion-dollar projects do not produce usable levels of power already (and they don't), scaling them up is not going to produce a realistically competitive power source.
Different ideas for hot fusion, like inertial confinement, are promising. Although they are no closer to a power-plant today, they haven't been explored as well. The only scalable fusion we have today is the H-bomb. There once was a proposal for generating fusion energy by exploding hydrogen bombs in a gigantic underground enclosure. Even considering the enormous size of the enclosure, this would probably be cheaper and easier to make work than a Tokamak. Of course, the security danger of using hydrogen bombs in day-to-day activities probably rules this idea out.
Geothermal energy is very similar to extracting the heat from an underground nuclear explosion. It is a potentially limitless source of energy, and it will not require an open-ended neverending research program.
In this context, one must always mention the great work of Pons and Fleischmann on cold fusion, which is completely separate. This is a set of different research directions and ideas, and it makes all this hot fusion nonsense moot. It is clear from Pons and Fleischmann's seminal, era-defining work, that it is possible to use atomic scale chemistry to get fusion. This is where the hot fusion money should have been diverted back in 1989, and some speculate that the prospect of losing money led the hot-fusion scientists to suppress Pons and Fleischmann's work.