Eric Schulte writes: > This test (test-ob/catches-all-references) is from commit c21692506d8, > which doesn't have anything to do with newlines (judging from the commit > message). > > To me the more natural behavior is to include the newline in the > expansion. Maybe we have discussed this before on list, and decided > stripping the newline was preferable, but I don't recall that > discussion.
I seem to have misremembered. Anyway, when you committed that test you apparently thought it was more natural to leave the trailing newline out. I'm not using such constructs myself, but changing this behaviour more than two years later carries a risk that people have documents that rely on it, for good or bad reasons. > Just because behavior ends up being encoded in a test doesn't > necessarily mean the behavior is correct. I think as test suites > attempt to protect the desired behavior they often end up also > protecting incidental behavior of the implementation at the time the > test was written. Well, a test test should proof the specification is implemented correctly. This is exactly why test methodology has evolved to unit tests, which mostly avoids testing for incident behaviour or emergent phenomena. Regards, Achim. -- +<[Q+ Matrix-12 WAVE#46+305 Neuron microQkb Andromeda XTk Blofeld]>+ Wavetables for the Terratec KOMPLEXER: http://Synth.Stromeko.net/Downloads.html#KomplexerWaves
