Build reproducible tarballs from git

Author: Rovanion

easybuild/tools/filetools.py

@@ -2707,11 +2707,15 @@ def get_source_tarball_from_git(filename, targetdir, git_config):
             for cmd in cmds:
                 run.run_cmd(cmd, log_all=True, simple=True, regexp=False, path=repo_name)
 
-    # create an archive and delete the git repo directory
-    if keep_git_dir:
-        tar_cmd = ['tar', 'cfvz', targetpath, repo_name]
-    else:
-        tar_cmd = ['tar', 'cfvz', targetpath, '--exclude', '.git', repo_name]
+    # When CentOS 7 is phased out and tar>1.28 is everywhere, replace find-sort-pipe with tar-flag
+    # '--sort=name' and place LC_ALL in front of tar. Also remove flags --null, --no-recursion, and
+    # --files-from - from the flags to tar. See https://reproducible-builds.org/docs/archives/
+    tar_cmd = ['find', repo_name, '-print0', '-path \'*/.git\' -prune' if not keep_git_dir else '', '|',
+               'LC_ALL=C', 'sort', '--zero-terminated', '|',
+               'GZIP=--no-name', 'tar', '--create', '--file', targetpath, '--no-recursion',
+               '--gzip', '--mtime="1970-01-01 00:00Z"', '--owner=0', '--group=0',
+               '--numeric-owner', '--format=gnu', '--null',
+               '--no-recursion', '--files-from -']
     run.run_cmd(' '.join(tar_cmd), log_all=True, simple=True, regexp=False)
 
     # cleanup (repo_name dir does not exist in dry run mode)

test/framework/filetools.py

@@ -2784,7 +2784,10 @@ def run_check():
         expected = '\n'.join([
             r'  running command "git clone --depth 1 --branch tag_for_tests %(git_repo)s"',
             r"  \(in .*/tmp.*\)",
-            r'  running command "tar cfvz .*/target/test.tar.gz --exclude .git testrepository"',
+            r'  running command "find testrepository -print0 -path \'*/.git\' -prune | LC_ALL=C sort --zero-terminated'
+            rf" | GZIP=--no-name tar --create --file {self.test_prefix}/target/test.tar.gz  --no-recursion"
+            r' --gzip --mtime="1970-01-01 00:00Z" --owner=0 --group=0 --numeric-owner --format=gnu'
+            r' --null --no-recursion --files-from -"',
             r"  \(in .*/tmp.*\)",
         ]) % git_repo
         run_check()
@@ -2793,7 +2796,10 @@ def run_check():
         expected = '\n'.join([
             r'  running command "git clone --depth 1 --branch tag_for_tests %(git_repo)s test123"',
             r"  \(in .*/tmp.*\)",
-            r'  running command "tar cfvz .*/target/test.tar.gz --exclude .git test123"',
+            r'  running command "find test123 -print0 -path \'*/.git\' -prune | LC_ALL=C sort --zero-terminated'
+            rf" | GZIP=--no-name tar --create --file {self.test_prefix}/target/test.tar.gz --no-recursion"
+            r' --gzip --mtime="1970-01-01 00:00Z" --owner=0 --group=0 --numeric-owner --format=gnu'
+            r' --null --no-recursion --files-from -"',
             r"  \(in .*/tmp.*\)",
         ]) % git_repo
         run_check()
@@ -2803,7 +2809,10 @@ def run_check():
         expected = '\n'.join([
             r'  running command "git clone --depth 1 --branch tag_for_tests --recursive %(git_repo)s"',
             r"  \(in .*/tmp.*\)",
-            r'  running command "tar cfvz .*/target/test.tar.gz --exclude .git testrepository"',
+            r'  running command "find testrepository -print0 -path \'*/.git\' -prune | LC_ALL=C sort --zero-terminated'
+            rf" | GZIP=--no-name tar --create --file {self.test_prefix}/target/test.tar.gz --no-recursion"
+            r' --gzip --mtime="1970-01-01 00:00Z" --owner=0 --group=0 --numeric-owner --format=gnu'
+            r' --null --no-recursion --files-from -"',
             r"  \(in .*/tmp.*\)",
         ]) % git_repo
         run_check()
@@ -2812,7 +2821,10 @@ def run_check():
         expected = '\n'.join([
             r'  running command "git clone --branch tag_for_tests --recursive %(git_repo)s"',
             r"  \(in .*/tmp.*\)",
-            r'  running command "tar cfvz .*/target/test.tar.gz testrepository"',
+            r'  running command "find testrepository -print0  | LC_ALL=C sort --zero-terminated | GZIP=--no-name tar'
+            rf" --create --file {self.test_prefix}/target/test.tar.gz --no-recursion --gzip"
+            r' --mtime="1970-01-01 00:00Z" --owner=0 --group=0 --numeric-owner --format=gnu --null --no-recursion'
+            r' --files-from -"',
             r"  \(in .*/tmp.*\)",
         ]) % git_repo
         run_check()
@@ -2825,7 +2837,10 @@ def run_check():
             r"  \(in .*/tmp.*\)",
             r'  running command "git checkout 8456f86 && git submodule update --init --recursive"',
             r"  \(in testrepository\)",
-            r'  running command "tar cfvz .*/target/test.tar.gz --exclude .git testrepository"',
+            r'  running command "find testrepository -print0 -path \'*/.git\' -prune | LC_ALL=C sort --zero-terminated'
+            rf" | GZIP=--no-name tar --create --file {self.test_prefix}/target/test.tar.gz --no-recursion"
+            r' --gzip --mtime="1970-01-01 00:00Z" --owner=0 --group=0 --numeric-owner --format=gnu'
+            r' --null --no-recursion --files-from -"',
             r"  \(in .*/tmp.*\)",
         ]) % git_repo
         run_check()
@@ -2836,7 +2851,10 @@ def run_check():
             r"  \(in .*/tmp.*\)",
             r'  running command "git checkout 8456f86"',
             r"  \(in testrepository\)",
-            r'  running command "tar cfvz .*/target/test.tar.gz --exclude .git testrepository"',
+            r'  running command "find testrepository -print0 -path \'*/.git\' -prune | LC_ALL=C sort --zero-terminated'
+            rf" | GZIP=--no-name tar --create --file {self.test_prefix}/target/test.tar.gz --no-recursion"
+            r' --gzip --mtime="1970-01-01 00:00Z" --owner=0 --group=0 --numeric-owner --format=gnu'
+            r' --null --no-recursion --files-from -"',
             r"  \(in .*/tmp.*\)",
         ]) % git_repo
         run_check()