CFL's are typically rated at 10,000 hours. This is based on long term life testing of the gasses and phosphors that are used. This is the same rating that you will find on conventional florescent lamps and cold cathode lamps found in many LCD display panels.
However, the dominant failure mechanism seems to be the electronics that drive the actual tube. These have variable life, and from my experience, many of them don't get close to 10,000 hours. Some of them don't make 100's of hours.
Further, running them from a dimmer or triac based switch, as found in IR activated outdoor lights, can materially degrade the lifetime of the electronics in the bulbs. They can fail from the first turn on or anytime thereafter. The waveforms generated by the electronic switches seems to be really hard on the CFL's electronics.
For any application NOT driven from a conventional light switch or a relay, I use standard incandescent bulbs.
The efficiency issue is real during warm up. However, they are still more efficient that a standard bulb if the amount of light produced is sufficient to get the job done. They don't draw more power during warmup, they just put out less light. As measured with a light meter, they increase about 2 EV values (4x) in light intensity from turn on to full brightness in an indoor environment. In a cold environment, they will not reach full intensity, but they will still consume relatively little power.
I just ran a quick test with a 60 watt conventional bulb and a "60" watt CFL. The conventional bulb drew 58 watts at a 99% indicated power factor. My spot light meter indicated an EV value of 17-2/3 (it's marked in 1/3 EV steps).
The CFL started out at an EV of 16, 11 watts and a 58% PF. After a two minute warm up, it had stabilized at an EV of 17-2/3, 12 watts and a 58% PF.
It appears to be doing what it is advertised to do at beginning of life. I just don't expect that it will be doing it that for 10,000 hours.