Re: [PATCH RFC] Add a config verbose option fetch and push

---

quoted

+fetch.verbose::
+	If true, it is the same as setting "-v" on the command line. If it is
+	false or not defined, git will use the command line parameters to
+	decide the verboseness of fetch.
+

Don't you usually use the configured option as the default, and 
then let the command line options override it (e.g., by specifying
--no-verbose).

//Peter

This patch fixes this objection.

 Documentation/config.txt |    7 +++++++
 builtin/fetch.c          |    7 +++++++
 builtin/push.c           |    7 +++++++
 3 files changed, 21 insertions(+), 0 deletions(-)

diff --git a/Documentation/config.txt b/Documentation/config.txt
index 39140ba..fc88d02 100644
--- a/Documentation/config.txt
+++ b/Documentation/config.txt

@@ -860,6 +860,9 @@ fetch.unpackLimit::
 	especially on slow filesystems.  If not set, the value of
 	`transfer.unpackLimit` is used instead.
 
+fetch.verbose::
+	If true, it is the same as setting "-v" on the command line.
+
 format.attach::
 	Enable multipart/mixed attachments as the default for
 	'format-patch'.  The value can also be a double quoted string

@@ -1495,6 +1498,10 @@ push.default::
 * `tracking` push the current branch to its upstream branch.
 * `current` push the current branch to a branch of the same name.
 
+push.verbose::
+	If true, it is the same as using the '-v' flag on the command
+	line.
+
 rebase.stat::
 	Whether to show a diffstat of what changed upstream since the last
 	rebase. False by default.

diff --git a/builtin/fetch.c b/builtin/fetch.c
index 8470850..f4832fe 100644
--- a/builtin/fetch.c
+++ b/builtin/fetch.c

@@ -885,6 +885,12 @@ static int fetch_one(struct remote *remote, int argc, const char **argv)
 	return exit_code;
 }
 
+static int git_fetch_verbose_config(const char *var,const char *value, void *dummy)
+{
+	if(!strcmp("fetch.verbose",var))
+		verbosity = git_config_maybe_bool(NULL,value);
+}
+
 int cmd_fetch(int argc, const char **argv, const char *prefix)
 {
 	int i;

@@ -897,6 +903,7 @@ int cmd_fetch(int argc, const char **argv, const char *prefix)
 	for (i = 1; i < argc; i++)
 		strbuf_addf(&default_rla, " %s", argv[i]);
 
+	git_config(git_fetch_verbose_config,NULL);
 	argc = parse_options(argc, argv, prefix,
 			     builtin_fetch_options, builtin_fetch_usage, 0);
 

diff --git a/builtin/push.c b/builtin/push.c
index f4358b9..e907b11 100644
--- a/builtin/push.c
+++ b/builtin/push.c

@@ -202,6 +202,12 @@ static int do_push(const char *repo, int flags)
 	return !!errs;
 }
 
+static int git_push_verbose_config(const char *var, const char *value, void *d)
+{
+	if(!strcmp("push.verbose",var))
+		verbosity = git_config_maybe_bool(NULL,value);
+}
+
 int cmd_push(int argc, const char **argv, const char *prefix)
 {
 	int flags = 0;

@@ -229,6 +235,7 @@ int cmd_push(int argc, const char **argv, const char *prefix)
 	};
 
 	git_config(git_default_config, NULL);
+	git_config(git_push_verbose_config, NULL);
 	argc = parse_options(argc, argv, prefix, options, push_usage, 0);
 
 	if (deleterefs && (tags || (flags & (TRANSPORT_PUSH_ALL | TRANSPORT_PUSH_MIRROR))))

-- 
1.7.1.97.gf85c7

Since Peter Kjellerstedt wanted --ff-only, and you want --verbose. I
wonder whether a better solution wouldn't be to farm this
functionality out to the config parser.

I.e. you'd do something like:

    static struct option builtin_fetch_options[] = {
        OPT__PROGRAM_NAME("fetch"), /* this is new */
	    OPT__VERBOSITY(&verbosity),
    	OPT_BOOLEAN(0, "all", &all,
	    	    "fetch from all remotes"),
        ...

And then in your .gitconfig:

    [fetch "option"]
        verbose = 1

Is there any reason not to add such a general facility?

Ævar Arnfjörð Bjarmason wrote:

Since Peter Kjellerstedt wanted --ff-only, and you want --verbose. I
wonder whether a better solution wouldn't be to farm this
functionality out to the config parser.

[...]

    [fetch "option"]
        verbose = 1

Is there any reason not to add such a general facility?

That would completely ruin the scriptability of almost all commands.
Imagine the user added the following options as default:

  add --edit
  checkout --patch
  cherry-pick --no-commit
  commit --amend
  pull --rebase

I'm sure you can find one option that changes the command in something
completely different *for every command*.

In fact even pull --ff-only has the same problem since it will refuse
the merge in cases where an unmodified pull would go through.

-- 
Thomas Rast
trast@{inf,student}.ethz.ch

On Sat, May 22, 2010 at 10:44, Thomas Rast [off-list ref] wrote:

Ævar Arnfjörð Bjarmason wrote:
That would completely ruin the scriptability of almost all commands.
Imagine the user added the following options as default:

 add --edit
 checkout --patch
 cherry-pick --no-commit
 commit --amend
 pull --rebase

I'm sure you can find one option that changes the command in something
completely different *for every command*.

Sure. But so would adding this as git-add to your $PATH:

    #!/bin/sh
    /usr/lib/git-core/git-add --edit $@

Git already has plenty of ways to shoot yourself in the foot. I don't
see how it's worse if that's done through some generalized facility
which results in less special-case code in individual tools.

Ævar Arnfjörð Bjarmason wrote:

On Sat, May 22, 2010 at 10:44, Thomas Rast [off-list ref] wrote:

quoted

Ævar Arnfjörð Bjarmason wrote:
That would completely ruin the scriptability of almost all commands.
Imagine the user added the following options as default:
 add --edit

[...]

quoted

I'm sure you can find one option that changes the command in something
completely different *for every command*.

Sure. But so would adding this as git-add to your $PATH:

    #!/bin/sh
    /usr/lib/git-core/git-add --edit $@

Two points:

* This way is not documented in git-config(1), as the proposed
  interface would have to be; hence, it is not "official".

* More importantly, it doesn't work; for builtins such as git-add, not
  even if you put it under the `git --exec-path` (yes, I've tested
  this).

Git already has plenty of ways to shoot yourself in the foot.

Can't argue with that.

-- 
Thomas Rast
trast@{inf,student}.ethz.ch

On Sat, May 22, 2010 at 13:34, Thomas Rast [off-list ref] wrote:

Ævar Arnfjörð Bjarmason wrote:

quoted

On Sat, May 22, 2010 at 10:44, Thomas Rast [off-list ref] wrote:

quoted

Ævar Arnfjörð Bjarmason wrote:
That would completely ruin the scriptability of almost all commands.
Imagine the user added the following options as default:
 add --edit

[...]

quoted

quoted

I'm sure you can find one option that changes the command in something
completely different *for every command*.

Sure. But so would adding this as git-add to your $PATH:

    #!/bin/sh
    /usr/lib/git-core/git-add --edit $@

Two points:

* This way is not documented in git-config(1), as the proposed
 interface would have to be; hence, it is not "official".

* More importantly, it doesn't work; for builtins such as git-add, not
 even if you put it under the `git --exec-path` (yes, I've tested
 this).

I actually tested it too and found that it didn't work. Then thought
"meh, it's pseudocode" and pressed "Send".

Aside from the specific implementation it's easy to make that work the
right way. You can alias the git command itself and munge its
arguments before passing them on to Git itself.

Anyway, this feature isn't something I actually care about. I only
wanted to suggest that if we're going to get lots of proposals to add
a specific config flag for some specific option in some specific
tool. That maybe it would be easier for everyone if there was some
general facility to do so. It would cut down on special-case code in
individual tools.

quoted

Git already has plenty of ways to shoot yourself in the foot.

Can't argue with that.

And perhaps a general facility might actually improve scriptability. A
determined user is going to override Git anyway, even if that means
some gross hack involving munging $PATH and overriding of Git
itself. At least if such users were pointed to a general facility
script writers could set GIT_IGNORE_CRAZY_USER_DIRECTIVES=1 or
something like that.