Merge pull request #3 from nbp/candidates/fetch-git

Optimize fetch-git.

src/lib/Hydra/Helper/AddBuilds.pm

@@ -315,7 +315,7 @@ sub fetchInputSystemBuild {
 sub fetchInputGit {
     my ($db, $project, $jobset, $name, $value) = @_;
 
-    (my $uri, my $branch) = split ' ', $value;
+    (my $uri, my $branch, my $deepClone) = split ' ', $value;
     $branch = defined $branch ? $branch : "master";
 
     my $timestamp = time;
@@ -325,42 +325,64 @@ sub fetchInputGit {
     mkpath(scmPath);
     my $clonePath = scmPath . "/" . sha256_hex($uri);
 
-    my $stdout; my $stderr;
+    my $stdout = ""; my $stderr = ""; my $res;
     if (! -d $clonePath) {
-        (my $res, $stdout, $stderr) = captureStdoutStderr(600,
+        # Clone everything and fetch the branch.
+        # TODO: Optimize the first clone by using "git init $clonePath" and "git remote add origin $uri".
+        ($res, $stdout, $stderr) = captureStdoutStderr(600,
             ("git", "clone", "--branch", $branch, $uri, $clonePath));
         die "Error cloning git repo at `$uri':\n$stderr" unless $res;
     }
 
-    # git pull + check rev
     chdir $clonePath or die $!; # !!! urgh, shouldn't do a chdir
-    (my $res, $stdout, $stderr) = captureStdoutStderr(600,
-        ("git", "pull", "--all"));
-    die "Error pulling latest change git repo at `$uri':\n$stderr" unless $res;
 
-    (my $res1, $stdout, $stderr) = captureStdoutStderr(600,
-        ("git", "ls-remote", $clonePath, $branch));
+    if (defined $deepClone) {
+        # This fetch every branches from the remote repository and create a
+        # local branch for each heads of the remote repository.  This is
+        # necessary to provide a working git-describe.
+        ($res, $stdout, $stderr) = captureStdoutStderr(600,
+            ("git", "pull", "--ff-only", "-fu", "--all", "origin"));
+        die "Error pulling latest change from git repo at `$uri':\n$stderr" unless $res;
+    } else {
+        # This command force the update of the local branch to be in the same as
+        # the remote branch for whatever the repository state is.  This command mirror
+        # only one branch of the remote repository.
+        ($res, $stdout, $stderr) = captureStdoutStderr(600,
+            ("git", "fetch", "-fu", "origin", "+$branch:$branch"));
+        die "Error fetching latest change from git repo at `$uri':\n$stderr" unless $res;
+    }
 
-    die "Cannot get head revision of Git branch '$branch' at `$uri':\n$stderr" unless $res1 ;
+    ($res, $stdout, $stderr) = captureStdoutStderr(600,
+        ("git", "rev-parse", "$branch"));
+    die "Error getting revision number of Git branch '$branch' at `$uri':\n$stderr" unless $res;
 
-    # Take the first commit ID returned by `ls-remote'.  The
-    # assumption is that `ls-remote' returned both `refs/heads/BRANCH'
-    # and `refs/remotes/origin/BRANCH', and that both point at the
-    # same commit.
-    my ($first) = split /\n/, $stdout;
-    (my $revision, my $ref) = split ' ', $first;
+    my ($revision) = split /\n/, $stdout;
     die unless $revision =~ /^[0-9a-fA-F]+$/;
+    die "Error getting a well-formated revision number of Git branch '$branch' at `$uri':\n$stdout" unless $res;
+
+    my $ref = "refs/heads/$branch";
+
+    # If deepClone is defined, then we look at the content of the repository
+    # to determine if this is a top-git branch.
+    if (defined $deepClone) {
+
+        # Checkout the branch to look at its content.
+        ($res, $stdout, $stderr) = captureStdoutStderr(600,
+            ("git", "checkout", "$branch"));
+        die "Error checking out Git branch '$branch' at `$uri':\n$stderr" unless $res;
 
         if (-f ".topdeps") {
             # This is a TopGit branch.  Fetch all the topic branches so
             # that builders can run "tg patch" and similar.
-        (my $res, $stdout, $stderr) = captureStdoutStderr(600,
+            ($res, $stdout, $stderr) = captureStdoutStderr(600,
                 ("tg", "remote", "--populate", "origin"));
 
             print STDERR "Warning: `tg remote --populate origin' failed:\n$stderr" unless $res;
         }
+    }
 
-    # Some simple caching: don't check a uri/branch more than once every hour, but prefer exact match on uri/branch/revision.
+    # Some simple caching: don't check a uri/branch/revision more than once.
+    # TODO: Fix case where the branch is reset to a previous commit.
     my $cachedInput ;
     ($cachedInput) = $db->resultset('CachedGitInputs')->search(
         {uri => $uri, branch => $branch, revision => $revision},
@@ -371,12 +393,14 @@ sub fetchInputGit {
         $sha256 = $cachedInput->sha256hash;
         $revision = $cachedInput->revision;
     } else {
-    
         # Then download this revision into the store.
-        print STDERR "checking out Git input from $uri\n";
+        print STDERR "checking out Git branch $branch from $uri\n";
         $ENV{"NIX_HASH_ALGO"} = "sha256";
         $ENV{"PRINT_PATH"} = "1";
+        $ENV{"NIX_PREFETCH_GIT_LEAVE_DOT_GIT"} = "0";
+        $ENV{"NIX_PREFETCH_GIT_DEEP_CLONE"} = "";
 
+        if (defined $deepClone) {
             # Checked out code often wants to be able to run `git
             # describe', e.g., code that uses Gnulib's `git-version-gen'
             # script.  Thus, we leave `.git' in there.  Same for
@@ -388,8 +412,9 @@ sub fetchInputGit {
             # http://thread.gmane.org/gmane.linux.distributions.nixos/3569
             # for a discussion.
             $ENV{"NIX_PREFETCH_GIT_DEEP_CLONE"} = "1";
+        }
 
-        (my $res, $stdout, $stderr) = captureStdoutStderr(600,
+        ($res, $stdout, $stderr) = captureStdoutStderr(600,
             ("nix-prefetch-git", $clonePath, $revision));
         die "Cannot check out Git repository branch '$branch' at `$uri':\n$stderr" unless $res;
 
@@ -509,7 +534,7 @@ sub fetchInputHg {
 
     # init local hg clone
 
-    my $stdout; my $stderr;
+    my $stdout = ""; my $stderr = "";
 
     mkpath(scmPath);
     my $clonePath = scmPath . "/" . sha256_hex($uri);
@@ -681,7 +706,7 @@ sub captureStdoutStderr {
 
     if ($@) {
         die unless $@ eq "timeout\n";   # propagate unexpected errors
-        return (undef, undef, undef);
+        return (undef, "", "timeout\n");
     } else {
         return ($res, $stdout, $stderr);
     }

src/script/nix-prefetch-git

@@ -63,8 +63,8 @@ Options:
       --url url       Any url understand by 'git clone'.
       --rev ref       Any sha1 or references (such as refs/heads/master)
       --hash h        Expected hash.
-      --deepClone     Clone submodules recursively.
-      --no-deepClone  Do not clone submodules.
+      --deepClone     Clone history until a tag is found as parent.
+      --no-deepClone  Clone the minimum history.
       --leave-dotGit  Keep the .git directories.
       --builder       Clone as fetchgit does, but url, rev, and out option are mandatory.
 "
@@ -117,7 +117,14 @@ checkout_ref(){
 	# allow "git describe" and similar tools to work.  See
 	# http://thread.gmane.org/gmane.linux.distributions.nixos/3569
 	# for a discussion.
+
+        # To make git describe works, we need to fetch all tags.
+        if git fetch -t ${builder:+--progress} --depth 1 origin; then
 	    return 1
+        else
+            # There is no tag, don't try to recover git-describe mechanism.
+            deepClone=false
+        fi
     fi
 
     if test -z "$ref"; then
@@ -209,7 +216,7 @@ clone_user_rev() {
     # Allow doing additional processing before .git removal
     eval "$NIX_PREFETCH_GIT_CHECKOUT_HOOK"
     if test -z "$leaveDotGit"; then
-	echo "removing \`.git'..." >&2
+	test -n "$QUIET" || echo "removing \`.git'..." >&2
         find $dir -name .git\* | xargs rm -rf
     fi
 }
@@ -248,7 +255,7 @@ else
 
       # Compute the hash.
       hash=$(nix-hash --type $hashType $hashFormat $tmpFile)
-      if ! test -n "$QUIET"; then echo "hash is $hash" >&2; fi
+      test -n "$QUIET" || echo "hash is $hash" >&2;
 
       # Add the downloaded file to the Nix store.
       finalPath=$(nix-store --add-fixed --recursive "$hashType" $tmpFile)
@@ -259,7 +266,7 @@ else
       fi
   fi
 
-  if ! test -n "$QUIET"; then echo "path is $finalPath" >&2; fi
+  test -n "$QUIET" || echo "path is $finalPath" >&2
 
   echo $hash
 

tests/Setup.pm

@@ -8,7 +8,7 @@ use Hydra::Helper::AddBuilds;
 use Cwd;
 
 our @ISA = qw(Exporter);
-our @EXPORT = qw(hydra_setup nrBuildsForJobset queuedBuildsForJobset nrQueuedBuildsForJobset createBaseJobset createJobsetWithOneInput evalSucceeds runBuild);
+our @EXPORT = qw(hydra_setup nrBuildsForJobset queuedBuildsForJobset nrQueuedBuildsForJobset createBaseJobset createJobsetWithOneInput evalSucceeds runBuild updateRepository);
 
 sub hydra_setup {
   my ($db) = @_;
@@ -74,4 +74,12 @@ sub runBuild {
   return captureStdoutStderr(60, ("../src/script/hydra-build", $build->id));
 }
 
+sub updateRepository {
+  my ($scm, $update, $repo) = @_;
+  my ($res, $stdout, $stderr) = captureStdoutStderr(60, ($update, $repo));
+  die "Unexpected update error with $scm: $stderr\n" unless $res;
+  print STDOUT "Update $scm repository: $stdout" if $stdout ne "";
+  return $stdout ne "";
+}
+
 1;

tests/evaluation-tests.pl

@@ -7,7 +7,7 @@ use Setup;
 
 my $db = Hydra::Model::DB->new;
 
-use Test::Simple tests => 28;
+use Test::Simple tests => 48;
 
 hydra_setup($db);
 
@@ -57,7 +57,17 @@ my @scminputs = ("svn", "svn-checkout", "git", "bzr", "bzr-checkout", "hg");
 foreach my $scm (@scminputs) {
   $jobset = createJobsetWithOneInput($scm, "$scm-input.nix", "src", $scm, "$jobsBaseUri/$scm-repo");
 
-  ok(evalSucceeds($jobset),                 "Evaluating jobs/$scm-input.nix should exit with return code 0.");
-  ok(nrQueuedBuildsForJobset($jobset) == 1, "Evaluating jobs/$scm-input.nix should result in 1 build in queue");
-}
+  my $c = 1;
+  my $q = 1;
+  do {
+    # Verify that it can be fetched and queued.
+    ok(evalSucceeds($jobset),                  "$c Evaluating jobs/$scm-input.nix should exit with return code 0."); $c++;
+    ok(nrQueuedBuildsForJobset($jobset) == $q, "$c Evaluating jobs/$scm-input.nix should result in 1 build in queue"); $c++;
 
+    # Verify that it is deterministic and not queued again.
+    ok(evalSucceeds($jobset),                  "$c Evaluating jobs/$scm-input.nix should exit with return code 0."); $c++;
+    ok(nrQueuedBuildsForJobset($jobset) == $q, "$c Evaluating jobs/$scm-input.nix should result in $q build in queue"); $c++;
+
+    $q++;
+  } while(updateRepository($scm, getcwd . "/jobs/$scm-update.sh", getcwd . "/$scm-repo/"));
+}

tests/jobs/bzr-checkout-update.sh

@@ -0,0 +1 @@
+#! /bin/sh

tests/jobs/bzr-update.sh

@@ -0,0 +1 @@
+#! /bin/sh

tests/jobs/git-update.sh

@@ -0,0 +1,26 @@
+#! /bin/sh
+
+cd "$1"
+STATE_FILE=.state
+if test -e $STATE_FILE; then
+    state=$(cat $STATE_FILE)
+else
+    state=0;
+fi
+
+case $state in
+    (0)
+    echo "Add new file."
+    touch git-file-2
+    git add git-file-2 >&2
+    git commit -m "add git file 2" git-file-2 >&2
+    ;;
+    (1)
+    echo "Rewrite commit."
+    echo 1 > git-file-2
+    git add git-file-2 >&2
+    git commit --amend -m "add git file 2" git-file-2 >&2
+    ;;
+esac
+
+echo $(($state + 1)) > $STATE_FILE

tests/jobs/hg-update.sh

@@ -0,0 +1 @@
+#! /bin/sh

tests/jobs/svn-checkout-update.sh

@@ -0,0 +1 @@
+#! /bin/sh

tests/jobs/svn-update.sh

@@ -0,0 +1 @@
+#! /bin/sh