Currently, each test sample has two files: We format x.ceylon and check if the result matches x.ceylon.formatted.
For almost all the tests I’ve recently written (where “recently” actually starts quite a while back – probably around 90%), those two files are identical, because if it takes several iterations until the test case is perfect, it’s easier to just keep the original formatted and copy it over instead of maintaining both an “ugly” and a “pretty” version separately.
This is of course a waste of effort – the test system should recognize when the x.ceylon.formatted file is missing, and then just use x.ceylon instead.
Currently, each test sample has two files: We format
x.ceylon
and check if the result matchesx.ceylon.formatted
.For almost all the tests I’ve recently written (where “recently” actually starts quite a while back – probably around 90%), those two files are identical, because if it takes several iterations until the test case is perfect, it’s easier to just keep the original formatted and copy it over instead of maintaining both an “ugly” and a “pretty” version separately.
This is of course a waste of effort – the test system should recognize when the
x.ceylon.formatted
file is missing, and then just usex.ceylon
instead.