On 03/26/2018 01:23 PM, Claes Redestad wrote: > Looks ok to me. > > Apart from the -source 1.9 -target 1.9 issue Paul already brought up, my only > concern is that the multiplicity of variants makes running this test somewhat > time > consuming[1]. Complete test coverage is important, so to cut down on testing > overheads then maybe it's time to examine if we can let go of some of the > optional > variants (BC_SB_SIZED, MH_SB_SIZED, ...)?
Maybe! I would prefer not to disable these in tests now, nor let go of some of them, because those strategies serve as backup options should the default strategy misbehave (like it does now, in some corner cases). Users do not really know which of those strategies are better tested, and which are not, so it seems sensible to test all of them. -Aleksey
