Re: [PATCH 1/3] configure: respect --without-curl flags

[Eric Sunshine <sunshine@sunshineco.com>]

On Wed, Mar 18, 2020 at 3:38 AM Đoàn Trần Công Danh
<congdanhqx@gmail.com> wrote:
> diff --git a/configure.ac b/configure.ac
> @@ -592,6 +592,9 @@ fi
>  # Define NO_CURL if you do not have libcurl installed.  git-http-pull and
>  # git-http-push are not built, and you cannot use http:// and https://
>  # transports.
>
> +# Respect --without-curl
> +if test "x$NO_CURL" != "xYesPlease"; then
> ...
>  if test -z "$NO_CURL"; then

I realize that GIT_PARSE_WITH() will either clear NO_CURL or set it to
literal "YesPlease", but the comment(s) in this file describing
NO_CURL say only to _define_ it to build without curl support. So, I'm
wondering if it would make more sense to take a looser view about the
value of NO_CURL. The existing check of NO_CURL doesn't bother looking
for a specific value, but instead just tests whether it has a value or
not. Perhaps the new check can be modeled after that one.

Also, I think you can reduce the scope of this change quite a bit by
merely wrapping the AC_CHECK_LIB() invocation. So, either:

    ...
    if test -z "$NO_CURL"; then
    GIT_STASH_FLAGS($CURLDIR)

    AC_CHECK_LIB([curl], [curl_global_init],
    [NO_CURL=],
    [NO_CURL=YesPlease])

    GIT_UNSTASH_FLAGS($CURLDIR)
    fi

    GIT_CONF_SUBST([NO_CURL])
    ...

or even:

    ...
    if test -z "$NO_CURL"; then
    AC_CHECK_LIB([curl], [curl_global_init],
    [NO_CURL=],
    [NO_CURL=YesPlease])
    fi
    ...

Same comments applies to the other patches, as well.