Thermonuclear supernovae result when interaction with a companion reignites nuclear fusion in a carbon-oxygen white dwarf, causing a thermonuclear runaway, a catastrophic gain in pressure, and the disintegration of the whole white dwarf. It is usually thought that fusion is reignited in near-pycnonuclear conditions when the white dwarf approaches the Chandrasekhar mass. I briefly describe two long-standing problems faced by this scenario, and our suggestion that these supernovae instead result from mergers of carbon-oxygen white dwarfs, including those that produce sub-Chandrasekhar mass remnants. I then turn to possible observational tests, in particular those that test the absence or presence of electron captures during the burning.