Web lists-archives.com

Re: [GSoC][PATCH] commit: add a commit.signOff config variable




On Sun, Feb 04 2018, Eric Sunshine jotted:

> --- >8 ---
> for cfg in true false
> do
>     for opt in '' --signoff --no-signoff
>     do
>         case "$opt:$cfg" in
>         --signoff:*|:true) expect= ;;
>         --no-signoff:*|:false) expect=! ;;
>         esac
>         test_expect_success "commit.signoff=$cfg & ${opt:---signoff omitted}" '
>             git -c commit.signoff=$cfg commit --allow-empty -m x $opt &&
>             eval "$expect git log -1 --format=%B | grep ^Signed-off-by:"
>         '
>     done
> done
> --- >8 ---
>
> A final consideration is that tests run slowly on Windows, and although
> it's nice to be thorough by testing all six combinations, you can
> probably exercise the new code sufficiently by instead testing just two
> combinations. For instance, instead of all six combinations, test just
> these two:
>
> --- >8 ---
> test_expect_success 'commit.signoff=true & --signoff omitted' '
>     git -c commit.signoff=true commit --allow-empty -m x &&
>     git log -1 --format=%B | grep ^Signed-off-by:
> '
>
> test_expect_success 'commit.signoff=true & --no-signoff' '
>     git -c commit.signoff=true commit --allow-empty -m x --no-signoff &&
>     ! git log -1 --format=%B | grep ^Signed-off-by:
> '
> --- >8 ---

I just skimmed this, but just to this question. I don't think we need to
worry about 2 v.s. 6 tests having an impact on Windows performance, it's
just massive amounts of tests like my in-flight wildmatch test series
that really matter.

But if we are worring about 2 v.s. 6 there's always my in-flight
EXPENSIVE_ON_WINDOWS prereq :)