[PATCH RESEND] Git.pm: Set GIT_WORK_TREE if we set GIT_DIR

From: Frank Lichtenheld <redacted>

Otherwise git will use the current directory as work tree which will
lead to unexpected results if we operate in sub directory of the
work tree.

Signed-off-by: Frank Lichtenheld <redacted>
---
 perl/Git.pm         |    2 ++
 t/t9700-perl-git.sh |    4 ++++
 t/t9700/test.pl     |   13 +++++++++++++
 3 files changed, 19 insertions(+), 0 deletions(-)

No comments and doesn't seem to have been applied, so resent unchanged.

diff --git a/perl/Git.pm b/perl/Git.pm
index 291ff5b..4313db7 100644
--- a/perl/Git.pm
+++ b/perl/Git.pm

@@ -1280,6 +1280,8 @@ sub _cmd_exec {
 	my ($self, @args) = @_;
 	if ($self) {
 		$self->repo_path() and $ENV{'GIT_DIR'} = $self->repo_path();
+		$self->repo_path() and $self->wc_path()
+			and $ENV{'GIT_WORK_TREE'} = $self->wc_path();
 		$self->wc_path() and chdir($self->wc_path());
 		$self->wc_subdir() and chdir($self->wc_subdir());
 	}

diff --git a/t/t9700-perl-git.sh b/t/t9700-perl-git.sh
index b4ca244..4eb7d3f 100755
--- a/t/t9700-perl-git.sh
+++ b/t/t9700-perl-git.sh

@@ -29,6 +29,10 @@ test_expect_success \
      git add . &&
      git commit -m "first commit" &&
 
+     echo "new file in subdir 2" > directory2/file2 &&
+     git add . &&
+     git commit -m "commit in directory2" &&
+
      echo "changed file 1" > file1 &&
      git commit -a -m "second commit" &&
 

diff --git a/t/t9700/test.pl b/t/t9700/test.pl
index 697daf3..d9b29ea 100755
--- a/t/t9700/test.pl
+++ b/t/t9700/test.pl

@@ -98,3 +98,16 @@ TODO: {
 	todo_skip 'config after wc_chdir', 1;
 	is($r->config("color.string"), "value", "config after wc_chdir");
 }
+
+# Object generation in sub directory
+chdir("directory2");
+my $r2 = Git->repository();
+is($r2->repo_path, $abs_repo_dir . "/.git", "repo_path (2)");
+is($r2->wc_path, $abs_repo_dir . "/", "wc_path (2)");
+is($r2->wc_subdir, "directory2/", "wc_subdir initial (2)");
+
+# commands in sub directory
+my $last_commit = $r2->command_oneline(qw(rev-parse --verify HEAD));
+like($last_commit, qr/^[0-9a-fA-F]{40}$/, 'rev-parse returned hash');
+my $dir_commit = $r2->command_oneline('log', '-n1', '--pretty=format:%H', '.');
+isnt($last_commit, $dir_commit, 'log . does not show last commit');

-- 
1.6.2.1

From: Frank Lichtenheld <redacted>

So far we only set it to absolute paths in some cases which lead
to problems like wc_chdir not working.

Signed-off-by: Frank Lichtenheld <redacted>
---
 perl/Git.pm     |    2 +-
 t/t9700/test.pl |   10 ++--------
 2 files changed, 3 insertions(+), 9 deletions(-)

Resent unchanged. There was one comment which I've reponded too and
argued that it didn't apply and there was no further objections.

diff --git a/perl/Git.pm b/perl/Git.pm
index 4313db7..e8df55d 100644
--- a/perl/Git.pm
+++ b/perl/Git.pm

@@ -185,7 +185,7 @@ sub repository {
 
 		if ($dir) {
 			$dir =~ m#^/# or $dir = $opts{Directory} . '/' . $dir;
-			$opts{Repository} = $dir;
+			$opts{Repository} = abs_path($dir);
 
 			# If --git-dir went ok, this shouldn't die either.
 			my $prefix = $search->command_oneline('rev-parse', '--show-prefix');

diff --git a/t/t9700/test.pl b/t/t9700/test.pl
index d9b29ea..6c70aec 100755
--- a/t/t9700/test.pl
+++ b/t/t9700/test.pl

@@ -86,18 +86,12 @@ close TEMPFILE;
 unlink $tmpfile;
 
 # paths
-is($r->repo_path, "./.git", "repo_path");
+is($r->repo_path, $abs_repo_dir . "/.git", "repo_path");
 is($r->wc_path, $abs_repo_dir . "/", "wc_path");
 is($r->wc_subdir, "", "wc_subdir initial");
 $r->wc_chdir("directory1");
 is($r->wc_subdir, "directory1", "wc_subdir after wc_chdir");
-TODO: {
-	local $TODO = "commands do not work after wc_chdir";
-	# Failure output is active even in non-verbose mode and thus
-	# annoying.  Hence we skip these tests as long as they fail.
-	todo_skip 'config after wc_chdir', 1;
-	is($r->config("color.string"), "value", "config after wc_chdir");
-}
+is($r->config("test.string"), "value", "config after wc_chdir");
 
 # Object generation in sub directory
 chdir("directory2");

-- 
1.6.2.1

On Thu, May 07, 2009 at 03:41:27PM +0200, Frank Lichtenheld wrote:

This looks obviously correct?

You could even skip the first chdir and use $self->wc_path() .
$self->wc_subdir() in the second one to save a syscall, I guess. ;-)

I've really forgot most of the code already so it's not worth much, but

Acked-by: Petr Baudis <redacted>

Frank Lichtenheld schrieb:

Unfortunately, this change breaks MinGW git because the absolute path that
this produces is MSYS-style /c/path/to/repo, but git does not understand
this; it should be c:/path/to/repo. This value is ultimately assigned to
GIT_DIR, but the path name mangling that usually happens when an MSYS
program (like perl) spawns a non-MSYS program (like git) does not happen.

Your commit message is quite vague about the problems that you have seen.
I vote to revert this change.

-- Hannes

Frank Lichtenheld schrieb:

On Mon, May 25, 2009 at 09:33:20AM +0200, Johannes Sixt wrote:

quoted

Frank Lichtenheld schrieb:

quoted

--- a/perl/Git.pm
+++ b/perl/Git.pm

@@ -185,7 +185,7 @@ sub repository {
 
 		if ($dir) {
 			$dir =~ m#^/# or $dir = $opts{Directory} . '/' . $dir;
-			$opts{Repository} = $dir;
+			$opts{Repository} = abs_path($dir);

Unfortunately, this change breaks MinGW git because the absolute path that
this produces is MSYS-style /c/path/to/repo, but git does not understand
this; it should be c:/path/to/repo. This value is ultimately assigned to
GIT_DIR, but the path name mangling that usually happens when an MSYS
program (like perl) spawns a non-MSYS program (like git) does not happen.

Your commit message is quite vague about the problems that you have seen.
I vote to revert this change.

Note that abs_path is already used twice in the same function. Why are those
usages not problematic? I would be happy to work with you on finding a patch
that doesn't break, but I have to admit that I have no idea of the
Windows<->Perl<->git interactions.

The result of abs_path() three lines below the cited context is never
passed to git; only its trailing part is ever used. This does not seem to
be problematic on Windows, according to the test suite.

The other use if abs_path() is about bare repositories and that is
certainly problematic, but nobody uses the tools written in perl in a bare
repository on Windows, obviously, otherwise we would have heard complaints. ;)

As for the problems, a part of the public API of the module simply doesn't work
(i.e. wc_chdir) which I fixed. If we can't fix it we should at least not pretend
that it works.

Since you keep repeating "does not work", without any specifics, I can't
help (and I'm not going to find out myself what "does not work").

-- Hannes

On Mon, May 25, 2009 at 09:33:20AM +0200, Johannes Sixt wrote:

Frank Lichtenheld schrieb:

quoted

From: Frank Lichtenheld <redacted>

So far we only set it to absolute paths in some cases which lead
to problems like wc_chdir not working.

Signed-off-by: Frank Lichtenheld <redacted>
---
 perl/Git.pm     |    2 +-
 t/t9700/test.pl |   10 ++--------
 2 files changed, 3 insertions(+), 9 deletions(-)

Resent unchanged. There was one comment which I've reponded too and
argued that it didn't apply and there was no further objections.

diff --git a/perl/Git.pm b/perl/Git.pm
index 4313db7..e8df55d 100644
--- a/perl/Git.pm
+++ b/perl/Git.pm

@@ -185,7 +185,7 @@ sub repository {
 
 		if ($dir) {
 			$dir =~ m#^/# or $dir = $opts{Directory} . '/' . $dir;
-			$opts{Repository} = $dir;
+			$opts{Repository} = abs_path($dir);

Unfortunately, this change breaks MinGW git because the absolute path that
this produces is MSYS-style /c/path/to/repo, but git does not understand
this; it should be c:/path/to/repo. This value is ultimately assigned to
GIT_DIR, but the path name mangling that usually happens when an MSYS
program (like perl) spawns a non-MSYS program (like git) does not happen.

Your commit message is quite vague about the problems that you have seen.
I vote to revert this change.

Note that abs_path is already used twice in the same function. Why are those
usages not problematic? I would be happy to work with you on finding a patch
that doesn't break, but I have to admit that I have no idea of the
Windows<->Perl<->git interactions.

As for the problems, a part of the public API of the module simply doesn't work
(i.e. wc_chdir) which I fixed. If we can't fix it we should at least not pretend
that it works.

Gruesse,
-- 
Frank Lichtenheld [off-list ref]
www: http://www.djpig.de/

On Wed, May 27, 2009 at 01:16:16PM +0200, Johannes Sixt wrote:

Frank Lichtenheld schrieb:

quoted

As for the problems, a part of the public API of the module simply doesn't work
(i.e. wc_chdir) which I fixed. If we can't fix it we should at least not pretend
that it works.

Since you keep repeating "does not work", without any specifics, I can't
help (and I'm not going to find out myself what "does not work").

Oh, sorry, I thought that the core problem would be obvious from the related test
suite changes. I can elaborate on that later this evening when I'm not at work.

Gruesse,
-- 
Frank Lichtenheld [off-list ref]
www: http://www.djpig.de/