Eric Sunshine [off-list ref] writes:
Since you've dropped the `--out` and `--err` options entirely, I
wonder if now would be a good time to shorten the function name, as
well, in order to further reduce the noise level. Since it now only
tests stdout (and doesn't deal with stderr), a name even shorter than
what I tried for the last version might be even better. So, for
instance, the name test_out_count() might not be too bad:
test_out_count = 0 git ls-files -o &&
"Test out" to me sound like trying something out and the part "out"
in the name no longer hints it is about "output"; you may have
shortened the name too much to be meaningful, I am afraid.
Is the helper used to check with anything but equality? Otherwise
you can lose "= " to make it shorter.
Having said all that, as an external interface, I wonder
test_line_count -e = 0 git-ls-files -o
would work better. It usually takes <op> <num> <file>, but when
$1 is a magic "-e", we shift it out and it becomes <op> <num> <cmd>...
On Sat, Jun 19, 2021 at 2:17 AM Junio C Hamano [off-list ref] wrote:
Eric Sunshine [off-list ref] writes:
quoted
what I tried for the last version might be even better. So, for
instance, the name test_out_count() might not be too bad:
test_out_count = 0 git ls-files -o &&
"Test out" to me sound like trying something out and the part "out"
in the name no longer hints it is about "output"; you may have
shortened the name too much to be meaningful, I am afraid.
Is the helper used to check with anything but equality? Otherwise
you can lose "= " to make it shorter.
Having said all that, as an external interface, I wonder
test_line_count -e = 0 git-ls-files -o
would work better. It usually takes <op> <num> <file>, but when
$1 is a magic "-e", we shift it out and it becomes <op> <num> <cmd>...
Indeed, I have no problem seeing this as a new mode of
test_line_count() triggered by an option. In fact, I suggested exactly
that[1] when this idea first arose (except I named the option `-c`
rather than `-e`, but the latter is fine). However, my suggestion was
pretty much shot down[2] (and I don't entirely disagree with [2],
which is why I didn't pursue the idea in [1]).
[1]: https://lore.kernel.org/git/CAPig+cS4tkXZLPDEWgEytzEOCR7oGrXyg1CZVKVPSXuJOifLjQ@mail.gmail.com/
[2]: https://lore.kernel.org/git/xmqq5z0fxlgn.fsf@gitster.g/