> Am 30.01.2020 um 23:10 schrieb Michael Orlitzky <mich...@orlitzky.com>: > > This is a good point, but it's a problem we already have. We don't make > any promises about the precise message contained in e.g. a ValueError, > but we still frequently test the contents. You just update it when you > change the message. The doctest is more of a sanity check here, telling > you that you've got two different answers and to go figure out which one > is right. If the code is right, you update the doctest.
I agree a doctest is easily updated when things change, but errors could easily be introduced as well. Ideally a test is written in a way that it does not test more than necessary – and the iteration order of dictionaries usually does not need to be tested. > The downside to sorting everything in the tests it that running the > tests already takes several hours (8+ here). How many CPU hours do we > collectively burn by sorting something 1293752893 times instead of > updating the test once? Although there are many tests outputting dictionaries, these are usually very small, so I do not expect this to have a big impact on running times. Besides, dictionaries are currently sorted via IPython during testing as well. -- You received this message because you are subscribed to the Google Groups "sage-devel" group. To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/sage-devel/5318CC36-7274-4849-9CDF-616405786930%40gmail.com.
