Junio C Hamano <gitster@xxxxxxxxx> writes: >>> @@ -146,7 +146,10 @@ test_expect_success 'non-hex character in RGB color' ' >>> invalid_color "#12x456" && >>> invalid_color "#123x56" && >>> invalid_color "#1234x6" && >>> - invalid_color "#12345x" >>> + invalid_color "#12345x" && >>> + invalid_color "#x23" && >>> + invalid_color "#1x3" && >>> + invalid_color "#12x" >>> ' >> >> This made me wonder what we'd do with "#1", "#12", "#1234", etc. Looking >> at the code change, I think we'd continue to reject them. I wonder if it >> is worth covering here. > > Worth covering in this test, yes, but I am perfectly OK with leaving > it outside the series as a #leftoverbit clean-up. Ah, I take it back. The preimage was added by [2/3] so it is fair to say that that step would be the right place to do that from the get-go.
