The latest report on the F-35 program by Michael Gilmore, the Pentagon’s director of operational test and evaluation (DOT&E), spotlights growing problems with late software deliveries for the stealthy fighter.

Software releases in 2012, the report says, ran late as compared to the schedule adopted after the 2010 technical baseline review, which was carried out in part to correct optimistic projections made before that date. (The program’s leaders had underestimated the amount of regression testing — tests to make sure that changes had not induced problems in previously tested functions — and overestimated test rates and productivity.)

Block 1 software is not complete. Lot 2 and Lot 3 aircraft have been delivered “with major variances against the expected capabilities,” the report says.

Block 2A, the initial training software, was four months late and less than half of it was available at the point where the report was written. Block 2B, intended to be the first combat-capable software, is late. Block 3i (interim), a bridge between 2B and the service-standard Block 3F, “has lagged in integration and laboratory testing.”

Software problems are part but not all of the reason for slow progress with weapons integration, along with optimistic and inaccurate assumptions about the need for margins and the availability of instrumentation and range support. “The impact of these delays will potentially require an additional 18 months added to the schedule for weapons integration events,” the report warns.

The report adds to the uncertainty surrounding the F-35’s initial operational capability (IOC) dates. Last summer, Congress added language to the 2013 budget that called on the U.S. Air Force and Navy to name IOC dates for all three versions by year’s end — then changed the deadline to June 1 at the last minute. The most recent Selected Acquisition Report disclosed that Block 3F initial operational test and evaluation, a necessary event for IOC, would not be finished until 2019 — and that does not include any additional weapons integration time.

Meanwhile, some earlier warnings from the DOT&E about the aircraft are proving prescient. DOT&E warned three years ago that the removal of check valves from fluid lines — which automatically stop the flow if the line is damaged downstream — would make the aircraft more vulnerable. That was before the start of live-fire testing at China Lake, Calif.

Now, testing has shown that the polyalphaolefin (PAO, related to synthetic motor oil) coolant in the avionics system poses a fire risk. “The threat in this ballistic test ruptured the PAO pressure line in the area just below the cockpit, causing a sustained PAO-based fire with a leak rate of 2.2 gallons per minute,” the report says. In engineering language: “a similar event in flight would likely cause an immediate incapacitation and loss of the pilot and aircraft.”

A fix for that problem is not impossible, although difficult. Most check systems are triggered by higher leak rates, so designing a valve that shuts off below a catastrophic fire level, but that does not result in high false-alarm rates, is tricky.