this means having to deal with escape codes. a lot of the use of regexp is to not have to worry about escape codes, not to deal with dynamic input. i don't see this as urgent technical debt; is there a particular place where regexp testing bit us?

a lot of the use of regexp is to not have to worry about escape codes, not to deal with dynamic input.

It used to be, but we never output ANSI escape codes in tests anymore, even those that are in "TTY" mode, since iostreams.Test() always returns an IOStreams instance that has color disabled, and the ColorScheme object respects that.

Even if we were outputting ANSI escape codes, I would still suggest that we do exact matching of output after we manually stripped them. So, it would be something like:

output = stripansi.Strip(stdout.String())
assert.Equal(t, "...", output)

But I'm not sure if there is a single place in the codebase where we would need to apply such an approach right now.

i don't see this as urgent technical debt; is there a particular place where regexp testing bit us?

No, it's not an urgent technical debt. I'm trying to get us to the point where we can drop the ambiguously named test package. A lot of what ExpectLines was used for an be just as easily achieved with exact matches or with assert.Contains().