Am 28.12.20 um 05:54 schrieb Seth House:
> The tool-specific functions are sometimes needed in scope earlier than
> when run_merge_tool is called.
You should answer why this change is needed. "are sometimes needed in
scope earlier" cannot be true, else we would have a bug that is fixed by
this change. But this isn't a bug fix, is it?
It would be ok to say "We are going to add another thing that we will
need before run_merge_tool; this is a preparation" or something.
Which brings me to another point: I do not see that something is added
to initialize_merge_tool in a later patch. You are only replacing
setup_tool calls by initialize_merge_tool, which forwards to setup_tool.
Why do we need a new function?
>
> Signed-off-by: Seth House <seth@xxxxxxxxx>
> ---
> Documentation/git-mergetool--lib.txt | 4 ++++
> git-difftool--helper.sh | 2 ++
> git-mergetool--lib.sh | 7 ++++---
> git-mergetool.sh | 2 ++
> 4 files changed, 12 insertions(+), 3 deletions(-)
>
> diff --git a/Documentation/git-mergetool--lib.txt b/Documentation/git-mergetool--lib.txt
> index 4da9d24096..3e8f59ac0e 100644
> --- a/Documentation/git-mergetool--lib.txt
> +++ b/Documentation/git-mergetool--lib.txt
> @@ -38,6 +38,10 @@ get_merge_tool_cmd::
> get_merge_tool_path::
> returns the custom path for a merge tool.
>
> +initialize_merge_tool::
> + bring merge tool specific functions into scope so they can be used or
> + overridden.
> +
> run_merge_tool::
> launches a merge tool given the tool name and a true/false
> flag to indicate whether a merge base is present.
[ swapped hunks for better sentence structure ]
> diff --git a/git-mergetool--lib.sh b/git-mergetool--lib.sh
> index 7225abd811..e059b3559e 100644
> --- a/git-mergetool--lib.sh
> +++ b/git-mergetool--lib.sh
> @@ -248,6 +248,10 @@ trust_exit_code () {
> fi
> }
>
> +initialize_merge_tool () {
> + # Bring tool-specific functions into scope
> + setup_tool "$1" || return 1
> +}
>
> # Entry point for running tools
> run_merge_tool () {
> @@ -259,9 +263,6 @@ run_merge_tool () {
> merge_tool_path=$(get_merge_tool_path "$1") || exit
> base_present="$2"
>
> - # Bring tool-specific functions into scope
> - setup_tool "$1" || return 1
> -
Before this change, run_merge_tool would exit here when there was an
error during setup_tool. But...
> diff --git a/git-difftool--helper.sh b/git-difftool--helper.sh
> index 46af3e60b7..c47a6d4253 100755
> --- a/git-difftool--helper.sh
> +++ b/git-difftool--helper.sh
> @@ -61,6 +61,7 @@ launch_merge_tool () {
> export BASE
> eval $GIT_DIFFTOOL_EXTCMD '"$LOCAL"' '"$REMOTE"'
> else
> + initialize_merge_tool "$merge_tool"
> run_merge_tool "$merge_tool"
> fi
... after the change we do not exit anymore. Does it matter?
Perhaps
initialize_merge_tool "$merge_tool" &&
run_merge_tool "$merge_tool"
> }
> @@ -79,6 +80,7 @@ if test -n "$GIT_DIFFTOOL_DIRDIFF"
> then
> LOCAL="$1"
> REMOTE="$2"
> + initialize_merge_tool "$merge_tool"
> run_merge_tool "$merge_tool" false
> else
> # Launch the merge tool on each path provided by 'git diff'
> diff --git a/git-mergetool.sh b/git-mergetool.sh
> index e3c7d78d1d..929192d0f8 100755
> --- a/git-mergetool.sh
> +++ b/git-mergetool.sh
> @@ -334,6 +334,8 @@ merge_file () {
> checkout_staged_file 2 "$MERGED" "$LOCAL"
> checkout_staged_file 3 "$MERGED" "$REMOTE"
>
> + initialize_merge_tool "$merge_tool"
> +
> if test "$(
> git config --get --bool "mergetool.$merge_tool.automerge" ||
> git config --get --bool "mergetool.automerge" ||
>
