And yet... Hard for me to believe that a broad-surface soldered joint like this one would offer structural flexibility sufficient to absorb the amount of distortion that would cause the windshield to pop out of its frame (which, we should remember, is lined with a compliant rubber seal). Not disputing Mr. Matsuo's claim, but I wonder whether the flexibility that he spoke of had more to do with positioning flexibility (compared to spot welding) for the two parts during and just after the joining process*, rather than after the car hit the road.
Put another way, just how much would that joint need to 'give' in order to prevent the amount of windshield frame distortion needed to make the windshield pop out: 1/16"? Is that soldered joint really capable of absorbing 1/16" of relative shear without breaking?