Đoàn Trần Công Danh <congdanhqx@xxxxxxxxx> writes: >> By the way, my opinion would be quite different if example like this >> one ... >> >> > test_expect_success 'tag --no-contains <existent_tag>' ' >> > - git tag --no-contains "v1.0" >actual 2>actual.err && >> > - test_line_count = 0 actual && >> > - test_line_count = 0 actual.err >> > + test_line_count_cmd --out = 0 --err = 0 git tag --no-contains v1.0 >> > ' >> >> ... were the majority, but I do not think that is the case. Most >> tests that employ the new test_line_count_cmd in this patch still >> create either actual or actual.err in the working tree anyway, so I >> do not see much point in adding this new helper---it is hard to >> explain to new test writers when to use it. > > I'm not sure if I get your opinion. Did you mean you wouldn't take > whole helper? Or you meant you still wanted to see a new helper for > checking only stdout? If it's the former, I'll send a different > series to only clean "git ls-files ... | wc -l" in t6400 and t6402, > if it's the latter, I'll rewrite without --err. I did not see much point in adding test_line_count_cmd with --out and/or --err options; the upside of having it was dubious after looking at the users of it in the patch that we are discussing. I did not consider test_line_count_cmd that only works on the standard output stream. From the patch under discussion, it is not immediately obvious how much such a simplified helper would help clean up the existing tests, so I have no opinion without seeing at least some sample uses.
