[PATCH 0/6] Minimal patches to let reftable pass the CI builds

[PATCH 0/6] Minimal patches to let reftable pass the CI builds

[Johannes Schindelin via GitGitGadget]

These patches fix the PR build failures at 
https://github.com/git/git/pull/847/checks?check_run_id=1460683728, and are
designed to be squashed into the 16 patches of the libreftable v3 patch
series as submitted at 
https://lore.kernel.org/git/pull.847.v3.git.git.1606419752.gitgitgadget@gmail.com/
.

A smaller form of the first patch was offered on the mailing list on May
4th: 
https://lore.kernel.org/git/ff60fde10919b6b8c54ecb8f38b710fac37624e3.1588599086.git.gitgitgadget@gmail.com/
. The next three patches were presented to the mailing list on October 2nd: 
https://lore.kernel.org/git/nycvar.QRO.7.76.6.2010021555290.50@tvgsbejvaqbjf.bet/
, and 
https://lore.kernel.org/git/nycvar.QRO.7.76.6.2010021557570.50@tvgsbejvaqbjf.bet/
. Some of the patches required slight adjustments to accommodate for context
changes.

Going forward, I would like to avoid the impression that it is the
responsibility of the Git for Windows maintainer to keep the CI build
passing on Windows. I am happy to assist in case it is unclear how to fix
certain issues. I am not happy having to implement and test those fixes
myself. How can we ensure this doesn't happen in the future?

Johannes Schindelin (6):
  fixup! reftable: rest of library
  fixup! reftable: utility functions
  fixup! reftable: rest of library
  fixup! reftable: rest of library
  fixup! reftable: rest of library
  fixup! reftable: rest of library

 contrib/buildsystems/CMakeLists.txt | 14 +++++++--
 reftable/dump.c                     | 36 ++++++++++-------------
 reftable/stack.c                    | 37 ++++++++++++++++++------
 reftable/stack_test.c               | 44 ++++++++++++++++++-----------
 4 files changed, 82 insertions(+), 49 deletions(-)


base-commit: 6229da992e43f4a2d6f4ccaccf2dbbdf11bd5a4a
Published-As: https://github.com/gitgitgadget/git/releases/tag/pr-801%2Fdscho%2Freftable-on-windows-v1
Fetch-It-Via: git fetch https://github.com/gitgitgadget/git pr-801/dscho/reftable-on-windows-v1
Pull-Request: https://github.com/gitgitgadget/git/pull/801
-- 
gitgitgadget

[PATCH 1/6] fixup! reftable: rest of library

[Johannes Schindelin via GitGitGadget]

From: Johannes Schindelin <johannes.schindelin@gmx.de>

Close the file descriptors to obsolete files before trying to delete or
rename them. This is actually required on Windows.

Note: this patch is just a band-aid to get the tests pass on Windows.
The fact that it is needed raises concerns about the overall resource
handling: are file descriptors closed properly whenever appropriate, or
are they closed much later (which can lead to rename() problems on
Windows, and risks running into ulimits)?

Also, a `reftable_stack_destroy()` call had to be moved in
`test_reftable_stack_uptodate()` to avoid the prompt complaining that a
`.ref` file could not be deleted on Windows. This raises the question
whether the code does the right thing when two concurrent processes want
to access the reftable, and one wants to compact it. At the moment, it
does not appear to fail gracefully.

Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
---
 reftable/stack.c      | 37 ++++++++++++++++++++++++++++---------
 reftable/stack_test.c |  2 +-
 2 files changed, 29 insertions(+), 10 deletions(-)

diff --git a/reftable/stack.c b/reftable/stack.c
index 1d632937d7..02c6a370ba 100644
--- a/reftable/stack.c
+++ b/reftable/stack.c
@@ -212,7 +212,6 @@ static int reftable_stack_reload_once(struct reftable_stack *st, char **names,
 		goto done;
 
 	new_tables = NULL;
-	st->readers_len = new_readers_len;
 	if (st->merged != NULL) {
 		merged_table_release(st->merged);
 		reftable_merged_table_free(st->merged);
@@ -220,6 +219,7 @@ static int reftable_stack_reload_once(struct reftable_stack *st, char **names,
 	if (st->readers != NULL) {
 		reftable_free(st->readers);
 	}
+	st->readers_len = new_readers_len;
 	st->readers = new_readers;
 	new_readers = NULL;
 	new_readers_len = 0;
@@ -939,14 +939,6 @@ static int stack_compact_range(struct reftable_stack *st, int first, int last,
 	strbuf_addstr(&new_table_path, "/");
 	strbuf_addbuf(&new_table_path, &new_table_name);
 
-	if (!is_empty_table) {
-		err = rename(temp_tab_file_name.buf, new_table_path.buf);
-		if (err < 0) {
-			err = REFTABLE_IO_ERROR;
-			goto done;
-		}
-	}
-
 	for (i = 0; i < first; i++) {
 		strbuf_addstr(&ref_list_contents, st->readers[i]->name);
 		strbuf_addstr(&ref_list_contents, "\n");
@@ -960,6 +952,32 @@ static int stack_compact_range(struct reftable_stack *st, int first, int last,
 		strbuf_addstr(&ref_list_contents, "\n");
 	}
 
+	/*
+	 * Now release the merged tables and readers
+	 */
+	if (st->merged != NULL) {
+		reftable_merged_table_free(st->merged);
+		st->merged = NULL;
+	}
+
+	if (st->readers != NULL) {
+		int i = 0;
+		for (i = 0; i < st->readers_len; i++) {
+			reader_close(st->readers[i]);
+			reftable_reader_free(st->readers[i]);
+		}
+		st->readers_len = 0;
+		FREE_AND_NULL(st->readers);
+	}
+
+	if (!is_empty_table) {
+		err = rename(temp_tab_file_name.buf, new_table_path.buf);
+		if (err < 0) {
+			err = REFTABLE_IO_ERROR;
+			goto done;
+		}
+	}
+
 	err = write(lock_file_fd, ref_list_contents.buf, ref_list_contents.len);
 	if (err < 0) {
 		err = REFTABLE_IO_ERROR;
@@ -1242,6 +1260,7 @@ static int stack_check_addition(struct reftable_stack *st,
 
 	free(refs);
 	reftable_iterator_destroy(&it);
+	reader_close(rd);
 	reftable_reader_free(rd);
 	return err;
 }
diff --git a/reftable/stack_test.c b/reftable/stack_test.c
index 11d3d30799..c35abd7301 100644
--- a/reftable/stack_test.c
+++ b/reftable/stack_test.c
@@ -159,12 +159,12 @@ static void test_reftable_stack_uptodate(void)
 	err = reftable_stack_add(st2, &write_test_ref, &ref2);
 	EXPECT(err == REFTABLE_LOCK_ERROR);
 
+	reftable_stack_destroy(st1);
 	err = reftable_stack_reload(st2);
 	EXPECT_ERR(err);
 
 	err = reftable_stack_add(st2, &write_test_ref, &ref2);
 	EXPECT_ERR(err);
-	reftable_stack_destroy(st1);
 	reftable_stack_destroy(st2);
 	clear_dir(dir);
 }
-- 
gitgitgadget

[PATCH 2/6] fixup! reftable: utility functions

[Johannes Schindelin via GitGitGadget]

From: Johannes Schindelin <johannes.schindelin@gmx.de>

Let's not forget our CMake configuration.

Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
---
 contrib/buildsystems/CMakeLists.txt | 14 ++++++++++++--
 1 file changed, 12 insertions(+), 2 deletions(-)

diff --git a/contrib/buildsystems/CMakeLists.txt b/contrib/buildsystems/CMakeLists.txt
index df539a44fa..f3a2fd3561 100644
--- a/contrib/buildsystems/CMakeLists.txt
+++ b/contrib/buildsystems/CMakeLists.txt
@@ -591,6 +591,12 @@ parse_makefile_for_sources(libxdiff_SOURCES "XDIFF_OBJS")
 list(TRANSFORM libxdiff_SOURCES PREPEND "${CMAKE_SOURCE_DIR}/")
 add_library(xdiff STATIC ${libxdiff_SOURCES})
 
+#reftable
+parse_makefile_for_sources(reftable_SOURCES "REFTABLE_OBJS")
+
+list(TRANSFORM reftable_SOURCES PREPEND "${CMAKE_SOURCE_DIR}/")
+add_library(reftable STATIC ${reftable_SOURCES})
+
 if(WIN32)
 	if(NOT MSVC)#use windres when compiling with gcc and clang
 		add_custom_command(OUTPUT ${CMAKE_BINARY_DIR}/git.res
@@ -613,7 +619,7 @@ endif()
 #link all required libraries to common-main
 add_library(common-main OBJECT ${CMAKE_SOURCE_DIR}/common-main.c)
 
-target_link_libraries(common-main libgit xdiff ${ZLIB_LIBRARIES})
+target_link_libraries(common-main libgit xdiff reftable ${ZLIB_LIBRARIES})
 if(Intl_FOUND)
 	target_link_libraries(common-main ${Intl_LIBRARIES})
 endif()
@@ -848,11 +854,15 @@ if(BUILD_TESTING)
 add_executable(test-fake-ssh ${CMAKE_SOURCE_DIR}/t/helper/test-fake-ssh.c)
 target_link_libraries(test-fake-ssh common-main)
 
+#reftable-tests
+parse_makefile_for_sources(test-reftable_SOURCES "REFTABLE_TEST_OBJS")
+list(TRANSFORM test-reftable_SOURCES PREPEND "${CMAKE_SOURCE_DIR}/")
+
 #test-tool
 parse_makefile_for_sources(test-tool_SOURCES "TEST_BUILTINS_OBJS")
 
 list(TRANSFORM test-tool_SOURCES PREPEND "${CMAKE_SOURCE_DIR}/t/helper/")
-add_executable(test-tool ${CMAKE_SOURCE_DIR}/t/helper/test-tool.c ${test-tool_SOURCES})
+add_executable(test-tool ${CMAKE_SOURCE_DIR}/t/helper/test-tool.c ${test-tool_SOURCES} ${test-reftable_SOURCES})
 target_link_libraries(test-tool common-main)
 
 set_target_properties(test-fake-ssh test-tool
-- 
gitgitgadget

[PATCH 3/6] fixup! reftable: rest of library

[Johannes Schindelin via GitGitGadget]

From: Johannes Schindelin <johannes.schindelin@gmx.de>

0-sized arrays are actually not portable.

Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
---
 reftable/stack_test.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/reftable/stack_test.c b/reftable/stack_test.c
index c35abd7301..cf2e32a25c 100644
--- a/reftable/stack_test.c
+++ b/reftable/stack_test.c
@@ -579,11 +579,11 @@ static void test_sizes_to_segments(void)
 
 static void test_sizes_to_segments_empty(void)
 {
-	uint64_t sizes[0];
+	uint64_t sizes[1];
 
 	int seglen = 0;
 	struct segment *segs =
-		sizes_to_segments(&seglen, sizes, ARRAY_SIZE(sizes));
+		sizes_to_segments(&seglen, sizes, 0);
 	EXPECT(seglen == 0);
 	reftable_free(segs);
 }
-- 
gitgitgadget

[PATCH 4/6] fixup! reftable: rest of library

[Johannes Schindelin via GitGitGadget]

From: Johannes Schindelin <johannes.schindelin@gmx.de>

Avoid using `getopt()`: it might be POSIX, but Git's audience is much
larger than POSIX. MSVC, for example, does not support `getopt()`.

Without this patch, the `vs-build` job of our CI/PR builds would simply
fail completely.

Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
---
 reftable/dump.c | 32 +++++++++++++-------------------
 1 file changed, 13 insertions(+), 19 deletions(-)

diff --git a/reftable/dump.c b/reftable/dump.c
index 0bd7a1cfa6..e990bffe0c 100644
--- a/reftable/dump.c
+++ b/reftable/dump.c
@@ -167,40 +167,34 @@ static void print_help(void)
 int reftable_dump_main(int argc, char *const *argv)
 {
 	int err = 0;
-	int opt;
 	int opt_dump_table = 0;
 	int opt_dump_stack = 0;
 	int opt_compact = 0;
-	const char *arg = NULL;
-	while ((opt = getopt(argc, argv, "2chts")) != -1) {
-		switch (opt) {
-		case '2':
-			hash_id = 0x73323536;
+	const char *arg = NULL, *argv0 = argv[0];
+
+	for (; argc > 1; argv++, argc--)
+		if (*argv[1] != '-')
 			break;
-		case 't':
+		else if (!strcmp("-2", argv[1]))
+			hash_id = 0x73323536;
+		else if (!strcmp("-t", argv[1]))
 			opt_dump_table = 1;
-			break;
-		case 's':
+		else if (!strcmp("-s", argv[1]))
 			opt_dump_stack = 1;
-			break;
-		case 'c':
+		else if (!strcmp("-c", argv[1]))
 			opt_compact = 1;
-			break;
-		case '?':
-		case 'h':
+		else if (!strcmp("-?", argv[1]) || !strcmp("-h", argv[1])) {
 			print_help();
 			return 2;
-			break;
 		}
-	}
 
-	if (argv[optind] == NULL) {
+	if (argc != 2) {
 		fprintf(stderr, "need argument\n");
 		print_help();
 		return 2;
 	}
 
-	arg = argv[optind];
+	arg = argv[1];
 
 	if (opt_dump_table) {
 		err = dump_table(arg);
@@ -211,7 +205,7 @@ int reftable_dump_main(int argc, char *const *argv)
 	}
 
 	if (err < 0) {
-		fprintf(stderr, "%s: %s: %s\n", argv[0], arg,
+		fprintf(stderr, "%s: %s: %s\n", argv0, arg,
 			reftable_error_str(err));
 		return 1;
 	}
-- 
gitgitgadget

[PATCH 5/6] fixup! reftable: rest of library

[Johannes Schindelin via GitGitGadget]

From: Johannes Schindelin <johannes.schindelin@gmx.de>

Yet another instance of `= {}` initialization.

Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
---
 reftable/dump.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/reftable/dump.c b/reftable/dump.c
index e990bffe0c..7d620a3cf0 100644
--- a/reftable/dump.c
+++ b/reftable/dump.c
@@ -82,7 +82,7 @@ static int dump_table(const char *tablename)
 static int compact_stack(const char *stackdir)
 {
 	struct reftable_stack *stack = NULL;
-	struct reftable_write_options cfg = {};
+	struct reftable_write_options cfg = { 0 };
 
 	int err = reftable_new_stack(&stack, stackdir, cfg);
 	if (err < 0)
@@ -101,7 +101,7 @@ static int compact_stack(const char *stackdir)
 static int dump_stack(const char *stackdir)
 {
 	struct reftable_stack *stack = NULL;
-	struct reftable_write_options cfg = {};
+	struct reftable_write_options cfg = { 0 };
 	struct reftable_iterator it = { NULL };
 	struct reftable_ref_record ref = { NULL };
 	struct reftable_log_record log = { NULL };
-- 
gitgitgadget

[PATCH 6/6] fixup! reftable: rest of library

[Johannes Schindelin via GitGitGadget]

From: Johannes Schindelin <johannes.schindelin@gmx.de>

The stack_test hard-codes `/tmp/`. That is a particular problem on
Windows where the temp directory is never at that location.

Let's not do that, but instead use `TMPDIR` as we do in similar
scenarios in the rest of Git's source code.

Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
---
 reftable/stack_test.c | 38 ++++++++++++++++++++++++--------------
 1 file changed, 24 insertions(+), 14 deletions(-)

diff --git a/reftable/stack_test.c b/reftable/stack_test.c
index cf2e32a25c..c9beaf4dbf 100644
--- a/reftable/stack_test.c
+++ b/reftable/stack_test.c
@@ -28,9 +28,19 @@ static void clear_dir(const char *dirname)
 	strbuf_release(&path);
 }
 
+static char *get_tmp_template(const char *prefix)
+{
+	static struct strbuf path = STRBUF_INIT;
+	const char *tmp = getenv("TMPDIR");
+
+	strbuf_reset(&path);
+	strbuf_addf(&path, "%s/%s.XXXXXX", tmp ? tmp : "/tmp", prefix);
+	return path.buf;
+}
+
 static void test_read_file(void)
 {
-	char fn[256] = "/tmp/stack.test_read_file.XXXXXX";
+	char *fn = get_tmp_template("stack.test_read_file");
 	int fd = mkstemp(fn);
 	char out[1024] = "line1\n\nline2\nline3";
 	int n, err;
@@ -99,7 +109,7 @@ static int write_test_log(struct reftable_writer *wr, void *arg)
 
 static void test_reftable_stack_add_one(void)
 {
-	char dir[256] = "/tmp/stack_test.XXXXXX";
+	char *dir = get_tmp_template("stack_test");
 	struct reftable_write_options cfg = { 0 };
 	struct reftable_stack *st = NULL;
 	int err;
@@ -132,7 +142,7 @@ static void test_reftable_stack_uptodate(void)
 	struct reftable_write_options cfg = { 0 };
 	struct reftable_stack *st1 = NULL;
 	struct reftable_stack *st2 = NULL;
-	char dir[256] = "/tmp/stack_test.XXXXXX";
+	char *dir = get_tmp_template("stack_test");
 	int err;
 	struct reftable_ref_record ref1 = {
 		.refname = "HEAD",
@@ -171,7 +181,7 @@ static void test_reftable_stack_uptodate(void)
 
 static void test_reftable_stack_transaction_api(void)
 {
-	char dir[256] = "/tmp/stack_test.XXXXXX";
+	char *dir = get_tmp_template("stack_test");
 	struct reftable_write_options cfg = { 0 };
 	struct reftable_stack *st = NULL;
 	int err;
@@ -216,7 +226,7 @@ static void test_reftable_stack_validate_refname(void)
 	struct reftable_write_options cfg = { 0 };
 	struct reftable_stack *st = NULL;
 	int err;
-	char dir[256] = "/tmp/stack_test.XXXXXX";
+	char *dir = get_tmp_template("stack_test");
 	int i;
 	struct reftable_ref_record ref = {
 		.refname = "a/b",
@@ -254,7 +264,7 @@ static int write_error(struct reftable_writer *wr, void *arg)
 
 static void test_reftable_stack_update_index_check(void)
 {
-	char dir[256] = "/tmp/stack_test.XXXXXX";
+	char *dir = get_tmp_template("stack_test");
 	struct reftable_write_options cfg = { 0 };
 	struct reftable_stack *st = NULL;
 	int err;
@@ -284,7 +294,7 @@ static void test_reftable_stack_update_index_check(void)
 
 static void test_reftable_stack_lock_failure(void)
 {
-	char dir[256] = "/tmp/stack_test.XXXXXX";
+	char *dir = get_tmp_template("stack_test");
 	struct reftable_write_options cfg = { 0 };
 	struct reftable_stack *st = NULL;
 	int err, i;
@@ -309,7 +319,7 @@ static void test_reftable_stack_add(void)
 		.exact_log_message = 1,
 	};
 	struct reftable_stack *st = NULL;
-	char dir[256] = "/tmp/stack_test.XXXXXX";
+	char *dir = get_tmp_template("stack_test");
 	struct reftable_ref_record refs[2] = { { NULL } };
 	struct reftable_log_record logs[2] = { { NULL } };
 	int N = ARRAY_SIZE(refs);
@@ -385,7 +395,7 @@ static void test_reftable_stack_log_normalize(void)
 		0,
 	};
 	struct reftable_stack *st = NULL;
-	char dir[256] = "/tmp/stack_test.XXXXXX";
+	char *dir = get_tmp_template("stack_test");
 
 	uint8_t h1[SHA1_SIZE] = { 0x01 }, h2[SHA1_SIZE] = { 0x02 };
 
@@ -436,7 +446,7 @@ static void test_reftable_stack_log_normalize(void)
 static void test_reftable_stack_tombstone(void)
 {
 	int i = 0;
-	char dir[256] = "/tmp/stack_test.XXXXXX";
+	char *dir = get_tmp_template("stack_test");
 	struct reftable_write_options cfg = { 0 };
 	struct reftable_stack *st = NULL;
 	int err;
@@ -511,7 +521,7 @@ static void test_reftable_stack_tombstone(void)
 
 static void test_reftable_stack_hash_id(void)
 {
-	char dir[256] = "/tmp/stack_test.XXXXXX";
+	char *dir = get_tmp_template("stack_test");
 	struct reftable_write_options cfg = { 0 };
 	struct reftable_stack *st = NULL;
 	int err;
@@ -621,7 +631,7 @@ static void test_suggest_compaction_segment_nothing(void)
 
 static void test_reflog_expire(void)
 {
-	char dir[256] = "/tmp/stack.test_reflog_expire.XXXXXX";
+	char *dir = get_tmp_template("stack.test_reflog_expire");
 	struct reftable_write_options cfg = { 0 };
 	struct reftable_stack *st = NULL;
 	struct reftable_log_record logs[20] = { { NULL } };
@@ -701,7 +711,7 @@ static void test_empty_add(void)
 	struct reftable_write_options cfg = { 0 };
 	struct reftable_stack *st = NULL;
 	int err;
-	char dir[256] = "/tmp/stack_test.XXXXXX";
+	char *dir = get_tmp_template("stack_test");
 	struct reftable_stack *st2 = NULL;
 
 	EXPECT(mkdtemp(dir));
@@ -723,7 +733,7 @@ static void test_reftable_stack_auto_compaction(void)
 {
 	struct reftable_write_options cfg = { 0 };
 	struct reftable_stack *st = NULL;
-	char dir[256] = "/tmp/stack_test.XXXXXX";
+	char *dir = get_tmp_template("stack_test");
 	int err, i;
 	int N = 100;
 	EXPECT(mkdtemp(dir));
-- 
gitgitgadget

Re: [PATCH 0/6] Minimal patches to let reftable pass the CI builds

[Han-Wen Nienhuys]

On Sat, Nov 28, 2020 at 7:44 AM Johannes Schindelin via GitGitGadget
<gitgitgadget@gmail.com> wrote:
>
> These patches fix the PR build failures at
> https://github.com/git/git/pull/847/checks?check_run_id=1460683728, and are
> designed to be squashed into the 16 patches of the libreftable v3 patch
> series as submitted at
> https://lore.kernel.org/git/pull.847.v3.git.git.1606419752.gitgitgadget@gmail.com/
> .
>
> A smaller form of the first patch was offered on the mailing list on May
> 4th:
> https://lore.kernel.org/git/ff60fde10919b6b8c54ecb8f38b710fac37624e3.1588599086.git.gitgitgadget@gmail.com/
> . The next three patches were presented to the mailing list on October 2nd:
> https://lore.kernel.org/git/nycvar.QRO.7.76.6.2010021555290.50@tvgsbejvaqbjf.bet/
> , and
> https://lore.kernel.org/git/nycvar.QRO.7.76.6.2010021557570.50@tvgsbejvaqbjf.bet/
> . Some of the patches required slight adjustments to accommodate for context
> changes.
>
> Going forward, I would like to avoid the impression that it is the
> responsibility of the Git for Windows maintainer to keep the CI build
> passing on Windows. I am happy to assist in case it is unclear how to fix
> certain issues. I am not happy having to implement and test those fixes
> myself. How can we ensure this doesn't happen in the future?

Thanks for the build system fix. I think it's OK to leave it to me to
solve the logic problems in the C code that you found.

Would you mind if I massaged these contributions directly back into
github.com/google/reftable? Google has a corporate CLA from Microsoft,
it's OK to accept this from you. For better or worse, this is still
where I'm developing reftable, until it has landed in git-core.

-- 
Han-Wen Nienhuys - Google Munich
I work 80%. Don't expect answers from me on Fridays.
--

Google Germany GmbH, Erika-Mann-Strasse 33, 80636 Munich

Registergericht und -nummer: Hamburg, HRB 86891

Sitz der Gesellschaft: Hamburg

Geschäftsführer: Paul Manicle, Halimah DeLaine Prado

Re: [PATCH 3/6] fixup! reftable: rest of library

[Jeff King]

On Sat, Nov 28, 2020 at 06:44:35AM +0000, Johannes Schindelin via GitGitGadget wrote:

> From: Johannes Schindelin <johannes.schindelin@gmx.de>
> 
> 0-sized arrays are actually not portable.

Definitely.

>  static void test_sizes_to_segments_empty(void)
>  {
> -	uint64_t sizes[0];
> +	uint64_t sizes[1];
>  
>  	int seglen = 0;
>  	struct segment *segs =
> -		sizes_to_segments(&seglen, sizes, ARRAY_SIZE(sizes));
> +		sizes_to_segments(&seglen, sizes, 0);
>  	EXPECT(seglen == 0);
>  	reftable_free(segs);

I think passing:

  sizes_to_segments(&seglen, NULL, 0);

may make the code more obvious. Unlike system functions like memcpy(),
we can be assured of whether our functions avoid looking at a
zero-length array (and size_to_segments does follow that rule).

  This function, of course, is nonsense that real code would not do, and
  is just unit-testing sizes_to_segments. I'm not wild in general about
  having a parallel suite of C tests that does not interact with our
  usual tests, but it may be the least bad way to benefit from the
  unit-test coverage that reftable ships with. As a rule, I'd much
  rather see a tool exposing functionality to the command-line, which
  can then be driven independently. I recognize that can end up
  complicated, though.

-Peff

Re: [PATCH 6/6] fixup! reftable: rest of library

[Jeff King]

On Sat, Nov 28, 2020 at 06:44:38AM +0000, Johannes Schindelin via GitGitGadget wrote:

> From: Johannes Schindelin <johannes.schindelin@gmx.de>
> 
> The stack_test hard-codes `/tmp/`. That is a particular problem on
> Windows where the temp directory is never at that location.
> 
> Let's not do that, but instead use `TMPDIR` as we do in similar
> scenarios in the rest of Git's source code.

Yeah, I noticed this, as well. This seems like a good band-aid, but it
would probably be nice if the test tool was able to write into a
specified directory (or even just the current directory). I don't see it
being invoked anywhere, but presumably if we were to add support to our
test suite, we'd have a script which invokes it within a scratch
directory.

-Peff

Re: [PATCH 3/6] fixup! reftable: rest of library

[Han-Wen Nienhuys]

On Tue, Dec 1, 2020 at 11:26 AM Jeff King <peff@peff.net> wrote:
>
> On Sat, Nov 28, 2020 at 06:44:35AM +0000, Johannes Schindelin via GitGitGadget wrote:
>
> > From: Johannes Schindelin <johannes.schindelin@gmx.de>
> >
> > 0-sized arrays are actually not portable.
>
> Definitely.
>
> >  static void test_sizes_to_segments_empty(void)
> >  {
> > -     uint64_t sizes[0];
> > +     uint64_t sizes[1];
> >
> >       int seglen = 0;
> >       struct segment *segs =
> > -             sizes_to_segments(&seglen, sizes, ARRAY_SIZE(sizes));
> > +             sizes_to_segments(&seglen, sizes, 0);
> >       EXPECT(seglen == 0);
> >       reftable_free(segs);
>
> I think passing:
>
>   sizes_to_segments(&seglen, NULL, 0);
>
> may make the code more obvious. Unlike system functions like memcpy(),
> we can be assured of whether our functions avoid looking at a
> zero-length array (and size_to_segments does follow that rule).
>
>   This function, of course, is nonsense that real code would not do, and

This test was added because 'real' code did this. In particular, if
you initialize a stack of reftables, the stack has zero elements. The
test was for the following bugfix

https://github.com/google/reftable/commit/b2e42ecb54e413e494c1fcc13c21e24422645007

Changing the array size to 1 here also prevents Valgrind to mark a
regression as a OOB write.

-- 
Han-Wen Nienhuys - Google Munich
I work 80%. Don't expect answers from me on Fridays.
--
Google Germany GmbH, Erika-Mann-Strasse 33, 80636 Munich
Registergericht und -nummer: Hamburg, HRB 86891
Sitz der Gesellschaft: Hamburg
Geschäftsführer: Paul Manicle, Halimah DeLaine Prado

Re: [PATCH 3/6] fixup! reftable: rest of library

[Jeff King]

On Tue, Dec 01, 2020 at 12:10:14PM +0100, Han-Wen Nienhuys wrote:

> > I think passing:
> >
> >   sizes_to_segments(&seglen, NULL, 0);
> >
> > may make the code more obvious. Unlike system functions like memcpy(),
> > we can be assured of whether our functions avoid looking at a
> > zero-length array (and size_to_segments does follow that rule).
> >
> >   This function, of course, is nonsense that real code would not do, and
> 
> This test was added because 'real' code did this. In particular, if
> you initialize a stack of reftables, the stack has zero elements. The
> test was for the following bugfix
> 
> https://github.com/google/reftable/commit/b2e42ecb54e413e494c1fcc13c21e24422645007
> 
> Changing the array size to 1 here also prevents Valgrind to mark a
> regression as a OOB write.

Sorry to be unclear. I meant that real code would not allocate a static
zero-length array (though it might do so on the heap). I agree that
making sure the function behaves in the 0-segment case is quite
reasonable.

Passing NULL actually makes the test more rigorous IMHO (because it is a
strict superset of failure modes, and is something an actual caller
might do if it were dynamically growing an array that started at NULL).

-Peff

Re: [PATCH 0/6] Minimal patches to let reftable pass the CI builds

[Johannes Schindelin]

Hi Han-Wen,

On Mon, 30 Nov 2020, Han-Wen Nienhuys wrote:

> On Sat, Nov 28, 2020 at 7:44 AM Johannes Schindelin via GitGitGadget
> <gitgitgadget@gmail.com> wrote:
> >
> > These patches fix the PR build failures at
> > https://github.com/git/git/pull/847/checks?check_run_id=1460683728, and are
> > designed to be squashed into the 16 patches of the libreftable v3 patch
> > series as submitted at
> > https://lore.kernel.org/git/pull.847.v3.git.git.1606419752.gitgitgadget@gmail.com/
> > .
> >
> > A smaller form of the first patch was offered on the mailing list on May
> > 4th:
> > https://lore.kernel.org/git/ff60fde10919b6b8c54ecb8f38b710fac37624e3.1588599086.git.gitgitgadget@gmail.com/
> > . The next three patches were presented to the mailing list on October 2nd:
> > https://lore.kernel.org/git/nycvar.QRO.7.76.6.2010021555290.50@tvgsbejvaqbjf.bet/
> > , and
> > https://lore.kernel.org/git/nycvar.QRO.7.76.6.2010021557570.50@tvgsbejvaqbjf.bet/
> > . Some of the patches required slight adjustments to accommodate for context
> > changes.
> >
> > Going forward, I would like to avoid the impression that it is the
> > responsibility of the Git for Windows maintainer to keep the CI build
> > passing on Windows. I am happy to assist in case it is unclear how to fix
> > certain issues. I am not happy having to implement and test those fixes
> > myself. How can we ensure this doesn't happen in the future?
>
> Thanks for the build system fix. I think it's OK to leave it to me to
> solve the logic problems in the C code that you found.
>
> Would you mind if I massaged these contributions directly back into
> github.com/google/reftable?

Whatever works best for you.

> Google has a corporate CLA from Microsoft, it's OK to accept this from
> you. For better or worse, this is still where I'm developing reftable,
> until it has landed in git-core.

Ciao,
Dscho

Re: [PATCH 6/6] fixup! reftable: rest of library

[Johannes Schindelin]

Hi Peff,

On Tue, 1 Dec 2020, Jeff King wrote:

> On Sat, Nov 28, 2020 at 06:44:38AM +0000, Johannes Schindelin via GitGitGadget wrote:
>
> > From: Johannes Schindelin <johannes.schindelin@gmx.de>
> >
> > The stack_test hard-codes `/tmp/`. That is a particular problem on
> > Windows where the temp directory is never at that location.
> >
> > Let's not do that, but instead use `TMPDIR` as we do in similar
> > scenarios in the rest of Git's source code.
>
> Yeah, I noticed this, as well. This seems like a good band-aid, but it
> would probably be nice if the test tool was able to write into a
> specified directory (or even just the current directory).

You can force it to do that by setting `TMPDIR` to the path of the current
directory...

> I don't see it being invoked anywhere,

It is invoked in `t/t0032-reftable-unittest.sh`:
https://github.com/dscho/git/blob/reftable-on-windows/t/t0032-reftable-unittest.sh

> but presumably if we were to add support to our test suite, we'd have a
> script which invokes it within a scratch directory.

I agree that it would make most sense for t0032 to prefix the call to
`test-tool` with `TMPDIR=$PWD`.

Ciao,
Dscho

Re: [PATCH 1/6] fixup! reftable: rest of library

[Han-Wen Nienhuys]

On Sat, Nov 28, 2020 at 7:44 AM Johannes Schindelin via GitGitGadget
<gitgitgadget@gmail.com> wrote:
>
> From: Johannes Schindelin <johannes.schindelin@gmx.de>
>
> Close the file descriptors to obsolete files before trying to delete or
> rename them. This is actually required on Windows.
>
> Note: this patch is just a band-aid to get the tests pass on Windows.
> The fact that it is needed raises concerns about the overall resource
> handling: are file descriptors closed properly whenever appropriate, or
> are they closed much later (which can lead to rename() problems on
> Windows, and risks running into ulimits)?
>
> Also, a `reftable_stack_destroy()` call had to be moved in
> `test_reftable_stack_uptodate()` to avoid the prompt complaining that a
> `.ref` file could not be deleted on Windows. This raises the question
> whether the code does the right thing when two concurrent processes want
> to access the reftable, and one wants to compact it. At the moment, it
> does not appear to fail gracefully.

Thanks for the report; I have to look more closely at these fixes; I
fear they might be incorrect.

The reftable spec doesn't treat this case in depth, and I think it was
rather written for Unix-like semantics. In the Unix flavor, a process
that wants to read can keep file descriptors open to keep reading from
the ref DB at a consistent snapshot.

What is the approach that the rest of Git on Windows takes in these
circumstances?

Consider processes P1 and P2, and the following sequence of actions

P1 opens ref DB (ie. opens a set of *.ref files for read)
P2 opens ref DB, executes a transaction. Post-transaction, it compacts
the reftable stack.
P2 exits
P1 exits

Currently, the compaction in P2 tries to delete the files obviated by
the compaction. On Windows this currently fails, because you can't
delete open files.

There are several options:

1) P2 should fail the compaction. This is bad because it will lead to
degraded performance over time, and it's not obvious if you can
anticipate that the deletion doesn't work.
2) P2 should retry deleting until it succeeds. This is bad, because a
reader can starve writers.
3) on exit, P1 should check if its *.ref files are still in use, and
delete them. This smells bad, because P1 is a read-only process, yet
it executes writes. Also, do we have good on-exit hooks in Git?
4) On exit, P1 does nothing. Stale *.ref files are left behind. Some
sort of GC process cleans things up asynchronously.
5) The ref DB should not keep files open, and should rather open and
close files as needed; this means P1 doesn't keep files open for long,
and P2 can retry safely.

I think 3) seems the cleanest to me (even though deleting in read
process feels weird), but perhaps we could fallback to 5) on windows
as well.

Thoughts?

What errno code does deleting an in-use file on Windows produce?


> Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
> ---
>  reftable/stack.c      | 37 ++++++++++++++++++++++++++++---------
>  reftable/stack_test.c |  2 +-
>  2 files changed, 29 insertions(+), 10 deletions(-)
>
> diff --git a/reftable/stack.c b/reftable/stack.c
> index 1d632937d7..02c6a370ba 100644
> --- a/reftable/stack.c
> +++ b/reftable/stack.c
> @@ -212,7 +212,6 @@ static int reftable_stack_reload_once(struct reftable_stack *st, char **names,
>                 goto done;
>
>         new_tables = NULL;
> -       st->readers_len = new_readers_len;
>         if (st->merged != NULL) {
>                 merged_table_release(st->merged);
>                 reftable_merged_table_free(st->merged);
> @@ -220,6 +219,7 @@ static int reftable_stack_reload_once(struct reftable_stack *st, char **names,
>         if (st->readers != NULL) {
>                 reftable_free(st->readers);
>         }
> +       st->readers_len = new_readers_len;
>         st->readers = new_readers;
>         new_readers = NULL;
>         new_readers_len = 0;
> @@ -939,14 +939,6 @@ static int stack_compact_range(struct reftable_stack *st, int first, int last,
>         strbuf_addstr(&new_table_path, "/");
>         strbuf_addbuf(&new_table_path, &new_table_name);
>
> -       if (!is_empty_table) {
> -               err = rename(temp_tab_file_name.buf, new_table_path.buf);
> -               if (err < 0) {
> -                       err = REFTABLE_IO_ERROR;
> -                       goto done;
> -               }
> -       }
> -
>         for (i = 0; i < first; i++) {
>                 strbuf_addstr(&ref_list_contents, st->readers[i]->name);
>                 strbuf_addstr(&ref_list_contents, "\n");
> @@ -960,6 +952,32 @@ static int stack_compact_range(struct reftable_stack *st, int first, int last,
>                 strbuf_addstr(&ref_list_contents, "\n");
>         }
>
> +       /*
> +        * Now release the merged tables and readers
> +        */
> +       if (st->merged != NULL) {
> +               reftable_merged_table_free(st->merged);
> +               st->merged = NULL;
> +       }
> +
> +       if (st->readers != NULL) {
> +               int i = 0;
> +               for (i = 0; i < st->readers_len; i++) {
> +                       reader_close(st->readers[i]);
> +                       reftable_reader_free(st->readers[i]);
> +               }
> +               st->readers_len = 0;
> +               FREE_AND_NULL(st->readers);
> +       }
> +
> +       if (!is_empty_table) {
> +               err = rename(temp_tab_file_name.buf, new_table_path.buf);
> +               if (err < 0) {
> +                       err = REFTABLE_IO_ERROR;
> +                       goto done;
> +               }
> +       }
> +
>         err = write(lock_file_fd, ref_list_contents.buf, ref_list_contents.len);
>         if (err < 0) {
>                 err = REFTABLE_IO_ERROR;
> @@ -1242,6 +1260,7 @@ static int stack_check_addition(struct reftable_stack *st,
>
>         free(refs);
>         reftable_iterator_destroy(&it);
> +       reader_close(rd);
>         reftable_reader_free(rd);
>         return err;
>  }
> diff --git a/reftable/stack_test.c b/reftable/stack_test.c
> index 11d3d30799..c35abd7301 100644
> --- a/reftable/stack_test.c
> +++ b/reftable/stack_test.c
> @@ -159,12 +159,12 @@ static void test_reftable_stack_uptodate(void)
>         err = reftable_stack_add(st2, &write_test_ref, &ref2);
>         EXPECT(err == REFTABLE_LOCK_ERROR);
>
> +       reftable_stack_destroy(st1);
>         err = reftable_stack_reload(st2);
>         EXPECT_ERR(err);
>
>         err = reftable_stack_add(st2, &write_test_ref, &ref2);
>         EXPECT_ERR(err);
> -       reftable_stack_destroy(st1);
>         reftable_stack_destroy(st2);
>         clear_dir(dir);
>  }
> --
> gitgitgadget
>


-- 
Han-Wen Nienhuys - Google Munich
I work 80%. Don't expect answers from me on Fridays.
--

Google Germany GmbH, Erika-Mann-Strasse 33, 80636 Munich

Registergericht und -nummer: Hamburg, HRB 86891

Sitz der Gesellschaft: Hamburg

Geschäftsführer: Paul Manicle, Halimah DeLaine Prado

Re: [PATCH 6/6] fixup! reftable: rest of library

[Jeff King]

On Tue, Dec 01, 2020 at 03:24:01PM +0100, Johannes Schindelin wrote:

> > I don't see it being invoked anywhere,
> 
> It is invoked in `t/t0032-reftable-unittest.sh`:
> https://github.com/dscho/git/blob/reftable-on-windows/t/t0032-reftable-unittest.sh

Ah, thank you. I was looking at what Junio has queued in hn/reftable.

> > but presumably if we were to add support to our test suite, we'd have a
> > script which invokes it within a scratch directory.
> 
> I agree that it would make most sense for t0032 to prefix the call to
> `test-tool` with `TMPDIR=$PWD`.

Yep, that would be fine.

I'm not sure if the long-term goal is to have this opaque unit-test
program or not. If it is, I was likewise going to suggest that its
ad-hoc output be replaced with TAP. But it looks like on your branch
that "test-tool reftable" does not produce output at all. So I may be a
bit behind on what the current state and forward plans are...

-Peff

Re: [PATCH 1/6] fixup! reftable: rest of library

[Johannes Schindelin]

Hi Han-Wen,

On Tue, 1 Dec 2020, Han-Wen Nienhuys wrote:

> On Sat, Nov 28, 2020 at 7:44 AM Johannes Schindelin via GitGitGadget
> <gitgitgadget@gmail.com> wrote:
> >
> > From: Johannes Schindelin <johannes.schindelin@gmx.de>
> >
> > Close the file descriptors to obsolete files before trying to delete or
> > rename them. This is actually required on Windows.
> >
> > Note: this patch is just a band-aid to get the tests pass on Windows.
> > The fact that it is needed raises concerns about the overall resource
> > handling: are file descriptors closed properly whenever appropriate, or
> > are they closed much later (which can lead to rename() problems on
> > Windows, and risks running into ulimits)?
> >
> > Also, a `reftable_stack_destroy()` call had to be moved in
> > `test_reftable_stack_uptodate()` to avoid the prompt complaining that a
> > `.ref` file could not be deleted on Windows. This raises the question
> > whether the code does the right thing when two concurrent processes want
> > to access the reftable, and one wants to compact it. At the moment, it
> > does not appear to fail gracefully.
>
> Thanks for the report; I have to look more closely at these fixes; I
> fear they might be incorrect.

They might be incorrect, but less so than the previous state, as testified
by the previously failing PR build.

> The reftable spec doesn't treat this case in depth, and I think it was
> rather written for Unix-like semantics. In the Unix flavor, a process
> that wants to read can keep file descriptors open to keep reading from
> the ref DB at a consistent snapshot.

Thanks for the explanation. I actually knew that.

> What is the approach that the rest of Git on Windows takes in these
> circumstances?

The rest of Git (whether on Windows or not) treats this as a no-no. You
cannot keep a handle open to a file that is deleted.

> Consider processes P1 and P2, and the following sequence of actions
>
> P1 opens ref DB (ie. opens a set of *.ref files for read)
> P2 opens ref DB, executes a transaction. Post-transaction, it compacts
> the reftable stack.
> P2 exits
> P1 exits
>
> Currently, the compaction in P2 tries to delete the files obviated by
> the compaction. On Windows this currently fails, because you can't
> delete open files.

Indeed. So the design needs to be fixed, if it fails.

> There are several options:
>
> 1) P2 should fail the compaction. This is bad because it will lead to
> degraded performance over time, and it's not obvious if you can
> anticipate that the deletion doesn't work.
> 2) P2 should retry deleting until it succeeds. This is bad, because a
> reader can starve writers.
> 3) on exit, P1 should check if its *.ref files are still in use, and
> delete them. This smells bad, because P1 is a read-only process, yet
> it executes writes. Also, do we have good on-exit hooks in Git?
> 4) On exit, P1 does nothing. Stale *.ref files are left behind. Some
> sort of GC process cleans things up asynchronously.
> 5) The ref DB should not keep files open, and should rather open and
> close files as needed; this means P1 doesn't keep files open for long,
> and P2 can retry safely.
>
> I think 3) seems the cleanest to me (even though deleting in read
> process feels weird), but perhaps we could fallback to 5) on windows
> as well.

Traditionally, Git would fail gracefully (i.e. with a warning) to delete
the stale files, and try again at a later stage (during `git gc --auto`,
for example, or after the next compaction step).

> What errno code does deleting an in-use file on Windows produce?

I believe it would be `EACCES`. See
https://docs.microsoft.com/en-us/cpp/c-runtime-library/reference/unlink-wunlink?view=msvc-160
for the documented behavior (I believe that an in-use file is treated the
same way as a read-only file in this instance).

Ciao,
Dscho

Re: [PATCH 6/6] fixup! reftable: rest of library

[Han-Wen Nienhuys]

On Wed, Dec 2, 2020 at 2:50 AM Jeff King <peff@peff.net> wrote:
> > > I don't see it being invoked anywhere,
> >
> > It is invoked in `t/t0032-reftable-unittest.sh`:
> > https://github.com/dscho/git/blob/reftable-on-windows/t/t0032-reftable-unittest.sh
>
> Ah, thank you. I was looking at what Junio has queued in hn/reftable.
>
> > > but presumably if we were to add support to our test suite, we'd have a
> > > script which invokes it within a scratch directory.
> >
> > I agree that it would make most sense for t0032 to prefix the call to
> > `test-tool` with `TMPDIR=$PWD`.
>
> Yep, that would be fine.
>
> I'm not sure if the long-term goal is to have this opaque unit-test
> program or not. If it is, I was likewise going to suggest that its
> ad-hoc output be replaced with TAP. But it looks like on your branch
> that "test-tool reftable" does not produce output at all. So I may be a
> bit behind on what the current state and forward plans are...

The most important requirement is that something fails if the
unittests don't work. I surmised that that meant running tests from
test-helper in some way, so this is what happens now. Looking for
"unit.?test" across the git codebase didn't turn up much info. Happy
to explore other solutions if you can give me pointers.

-- 
Han-Wen Nienhuys - Google Munich
I work 80%. Don't expect answers from me on Fridays.
--

Google Germany GmbH, Erika-Mann-Strasse 33, 80636 Munich
Registergericht und -nummer: Hamburg, HRB 86891
Sitz der Gesellschaft: Hamburg
Geschäftsführer: Paul Manicle, Halimah DeLaine Prado

Re: [PATCH 6/6] fixup! reftable: rest of library

[Jeff King]

On Wed, Dec 02, 2020 at 12:01:49PM +0100, Han-Wen Nienhuys wrote:

> > I'm not sure if the long-term goal is to have this opaque unit-test
> > program or not. If it is, I was likewise going to suggest that its
> > ad-hoc output be replaced with TAP. But it looks like on your branch
> > that "test-tool reftable" does not produce output at all. So I may be a
> > bit behind on what the current state and forward plans are...
> 
> The most important requirement is that something fails if the
> unittests don't work. I surmised that that meant running tests from
> test-helper in some way, so this is what happens now. Looking for
> "unit.?test" across the git codebase didn't turn up much info. Happy
> to explore other solutions if you can give me pointers.

Normally we do a combination of:

  - git plumbing exposes scriptable commands, and we make sure those
    work. This is a much coarser-grained unit than testing individual C
    functions, but produces resilient tests because that interface is
    user-visible and stable (and in fact seeing test breakages is often
    a sign that one will be breaking real users).

    These are obviously driven by shell script tests emulating what
    users might run.

  - for unit tests of individual data types where it's not appropriate
    to have a user-facing command, we often add a test-tool helper that
    exposes specific functions (e.g., t/helper/test-date.c exposes date
    parsing and formatting routines, test-oid-array.c exercises methods
    of that data type, etc).

    The C parts of these are usually generic, and then are driven by
    shell scripts providing the actual data in individual tests (and
    handling success/failure, skipping tests, etc).

    This is still coarser than you might get unit-testing inside C.
    E.g., you could not generally check the difference between passing
    an empty array vs NULL to a function. But our philosophy has
    generally been _not_ to test at that level. The C interfaces are
    internal, and Git is not that big a project. If there's a function
    whose caller does something unexpected, it's usually easier to fix
    the caller and add a test that triggers the caller's code path.

I agree that a test that simply runs a bunch of C code and either exits
with failure or success is better than nothing, in the sense of finding
tests. And wrapping that with a single test_expect_success does that.
But it's unfortunate that we get none of the fine-grained control that
the test suite provides, nor much support in debugging failing tests.

One middle ground would be for a battery of C tests to output
TAP-compatible output (outputting "ok 1 - foo works", and "not ok 2 -
bar does not work", etc). That at least gives more info on what fails,
and does it in a way that the rest of the test suite can interpret
(though not manipulate).

-Peff

Reftable locking on Windows (Re: [PATCH 1/6] fixup! reftable: rest of library)

[Han-Wen Nienhuys]

On Wed, Dec 2, 2020 at 1:32 PM Johannes Schindelin
<Johannes.Schindelin@gmx.de> wrote:
> > Consider processes P1 and P2, and the following sequence of actions
> >
> > P1 opens ref DB (ie. opens a set of *.ref files for read)
> > P2 opens ref DB, executes a transaction. Post-transaction, it compacts
> > the reftable stack.
> > P2 exits
> > P1 exits
> >
> > Currently, the compaction in P2 tries to delete the files obviated by
> > the compaction. On Windows this currently fails, because you can't
> > delete open files.
>
> Indeed. So the design needs to be fixed, if it fails.
>
> > There are several options:
> >
> > 1) P2 should fail the compaction. This is bad because it will lead to
> > degraded performance over time, and it's not obvious if you can
> > anticipate that the deletion doesn't work.
> > 2) P2 should retry deleting until it succeeds. This is bad, because a
> > reader can starve writers.
> > 3) on exit, P1 should check if its *.ref files are still in use, and
> > delete them. This smells bad, because P1 is a read-only process, yet
> > it executes writes. Also, do we have good on-exit hooks in Git?
> > 4) On exit, P1 does nothing. Stale *.ref files are left behind. Some
> > sort of GC process cleans things up asynchronously.
> > 5) The ref DB should not keep files open, and should rather open and
> > close files as needed; this means P1 doesn't keep files open for long,
> > and P2 can retry safely.
> >
> > I think 3) seems the cleanest to me (even though deleting in read
> > process feels weird), but perhaps we could fallback to 5) on windows
> > as well.
>
> Traditionally, Git would fail gracefully (i.e. with a warning) to delete
> the stale files, and try again at a later stage (during `git gc --auto`,
> for example, or after the next compaction step).

So, how does this sound:

* add a void

  set_warn_handler(void(*handler)(char const*))

The handler is called for non-fatal errors (eg. deleting an in-use
.ref file during compaction), and can provide the warning.

* all .ref files will be prefixed with an 8-byte random string, to
avoid that new *.ref files collide with unreferenced, stale *.ref
files.

* in reftable_stack_close(), after closing files, we check if *.ref
files we were reading have gone out of use. If so, delete those .ref
files, printing a warning on EACCESS.

* an on-exit handler in git calls reftable_stack_close()

* provide a reftable_stack_gc(), which loads tables.list and then
deletes all unused *.ref files that are older than tables.list.

> > What errno code does deleting an in-use file on Windows produce?
>
> I believe it would be `EACCES`. See
> https://docs.microsoft.com/en-us/cpp/c-runtime-library/reference/unlink-wunlink?view=msvc-160
> for the documented behavior (I believe that an in-use file is treated the
> same way as a read-only file in this instance).

your commit message says "to avoid the prompt complaining that a
`.ref` file could not be deleted on Windows." AFAICT, this prompt is
coming from windows itself, because the reftable library doesn't issue
this type of warning. Is there a way to delete a file that just
returns EACCESS if it fails, without triggering a prompt? Or is the
prompt part of the "failing gracefully" behavior?

-- 
Han-Wen Nienhuys - Google Munich
I work 80%. Don't expect answers from me on Fridays.
--
Google Germany GmbH, Erika-Mann-Strasse 33, 80636 Munich
Registergericht und -nummer: Hamburg, HRB 86891
Sitz der Gesellschaft: Hamburg
Geschäftsführer: Paul Manicle, Halimah DeLaine Prado

Re: Reftable locking on Windows (Re: [PATCH 1/6] fixup! reftable: rest of library)

[Ævar Arnfjörð Bjarmason]

On Wed, Dec 02 2020, Han-Wen Nienhuys wrote:

> On Wed, Dec 2, 2020 at 1:32 PM Johannes Schindelin
> <Johannes.Schindelin@gmx.de> wrote:
>> > Consider processes P1 and P2, and the following sequence of actions
>> >
>> > P1 opens ref DB (ie. opens a set of *.ref files for read)
>> > P2 opens ref DB, executes a transaction. Post-transaction, it compacts
>> > the reftable stack.
>> > P2 exits
>> > P1 exits
>> >
>> > Currently, the compaction in P2 tries to delete the files obviated by
>> > the compaction. On Windows this currently fails, because you can't
>> > delete open files.
>>
>> Indeed. So the design needs to be fixed, if it fails.
>>
>> > There are several options:
>> >
>> > 1) P2 should fail the compaction. This is bad because it will lead to
>> > degraded performance over time, and it's not obvious if you can
>> > anticipate that the deletion doesn't work.
>> > 2) P2 should retry deleting until it succeeds. This is bad, because a
>> > reader can starve writers.
>> > 3) on exit, P1 should check if its *.ref files are still in use, and
>> > delete them. This smells bad, because P1 is a read-only process, yet
>> > it executes writes. Also, do we have good on-exit hooks in Git?
>> > 4) On exit, P1 does nothing. Stale *.ref files are left behind. Some
>> > sort of GC process cleans things up asynchronously.
>> > 5) The ref DB should not keep files open, and should rather open and
>> > close files as needed; this means P1 doesn't keep files open for long,
>> > and P2 can retry safely.
>> >
>> > I think 3) seems the cleanest to me (even though deleting in read
>> > process feels weird), but perhaps we could fallback to 5) on windows
>> > as well.
>>
>> Traditionally, Git would fail gracefully (i.e. with a warning) to delete
>> the stale files, and try again at a later stage (during `git gc --auto`,
>> for example, or after the next compaction step).
>
> So, how does this sound:
>
> * add a void
>
>   set_warn_handler(void(*handler)(char const*))
>
> The handler is called for non-fatal errors (eg. deleting an in-use
> .ref file during compaction), and can provide the warning.
>
> * all .ref files will be prefixed with an 8-byte random string, to
> avoid that new *.ref files collide with unreferenced, stale *.ref
> files.

Just trying to follow along here. Do you mean prefix the file name or
the content of those *.ref files? In any case isn't this synonymous with
proposing moving beyond reftable v1 (to the WIP v2, or a v1.1 with just
this change?). The current spec seems to preclude prefixing either the
file content or filename, but maybe I'm misreading it:

https://googlers.googlesource.com/sop/jgit/+/reftable/Documentation/technical/reftable.md#update-transactions
https://googlers.googlesource.com/sop/jgit/+/reftable/Documentation/technical/reftable.md#header

When the reftable format was initially being discussed I suggested
somewhat flippantly (why not SQLite?). Partially because in trying to
invent what's essentially a portable and concurrently updated database
format we'd be likely to reinvent much of the same wheel.

So not to drag that whole thing up again as a proposal for a format
replacement, but I wonder what SQLite would do in this scenario & others
where desired DB semantics / transactions and FS/portability semantics
clash. AFAIK the reftable has only been battle-tested in production
under JGit so far (presumably Linux-only). Whereas SQLite has been
ported to and widely used on Windows, HP/UX and probably other systems
where Linux-specific assumptions don't apply.

> * in reftable_stack_close(), after closing files, we check if *.ref
> files we were reading have gone out of use. If so, delete those .ref
> files, printing a warning on EACCESS.
>
> * an on-exit handler in git calls reftable_stack_close()
>
> * provide a reftable_stack_gc(), which loads tables.list and then
> deletes all unused *.ref files that are older than tables.list.
>
>> > What errno code does deleting an in-use file on Windows produce?
>>
>> I believe it would be `EACCES`. See
>> https://docs.microsoft.com/en-us/cpp/c-runtime-library/reference/unlink-wunlink?view=msvc-160
>> for the documented behavior (I believe that an in-use file is treated the
>> same way as a read-only file in this instance).
>
> your commit message says "to avoid the prompt complaining that a
> `.ref` file could not be deleted on Windows." AFAICT, this prompt is
> coming from windows itself, because the reftable library doesn't issue
> this type of warning. Is there a way to delete a file that just
> returns EACCESS if it fails, without triggering a prompt? Or is the
> prompt part of the "failing gracefully" behavior?
>
> -- 
> Han-Wen Nienhuys - Google Munich
> I work 80%. Don't expect answers from me on Fridays.

Re: Reftable locking on Windows (Re: [PATCH 1/6] fixup! reftable: rest of library)

[Han-Wen Nienhuys]

On Thu, Dec 3, 2020 at 1:24 PM Ævar Arnfjörð Bjarmason <avarab@gmail.com> wrote:
> >> Indeed. So the design needs to be fixed, if it fails.
> >>
> >> > There are several options:
> >> >
> >> > 1) P2 should fail the compaction. This is bad because it will lead to
> >> > degraded performance over time, and it's not obvious if you can
> >> > anticipate that the deletion doesn't work.
> >> > 2) P2 should retry deleting until it succeeds. This is bad, because a
> >> > reader can starve writers.
> >> > 3) on exit, P1 should check if its *.ref files are still in use, and
> >> > delete them. This smells bad, because P1 is a read-only process, yet
> >> > it executes writes. Also, do we have good on-exit hooks in Git?
> >> > 4) On exit, P1 does nothing. Stale *.ref files are left behind. Some
> >> > sort of GC process cleans things up asynchronously.
> >> > 5) The ref DB should not keep files open, and should rather open and
> >> > close files as needed; this means P1 doesn't keep files open for long,
> >> > and P2 can retry safely.
> >> >
> >> > I think 3) seems the cleanest to me (even though deleting in read
> >> > process feels weird), but perhaps we could fallback to 5) on windows
> >> > as well.
> >>
> >> Traditionally, Git would fail gracefully (i.e. with a warning) to delete
> >> the stale files, and try again at a later stage (during `git gc --auto`,
> >> for example, or after the next compaction step).
> >
> > So, how does this sound:
> >
> > * add a void
> >
> >   set_warn_handler(void(*handler)(char const*))
> >
> > The handler is called for non-fatal errors (eg. deleting an in-use
> > .ref file during compaction), and can provide the warning.
> >
> > * all .ref files will be prefixed with an 8-byte random string, to
> > avoid that new *.ref files collide with unreferenced, stale *.ref
> > files.
>
> Just trying to follow along here. Do you mean prefix the file name or
> the content of those *.ref files? In any case isn't this synonymous with
> proposing moving beyond reftable v1 (to the WIP v2, or a v1.1 with just
> this change?). The current spec seems to preclude prefixing either the
> file content or filename, but maybe I'm misreading it:

This is about prefixing the filename with an uniquifying prefix. This
is because the range of timestamps encompassed by the table doesn't
uniquely identify the table.

Consider the following case:
  State: 1 table "00001.ref"
  Transaction 1: add ref xyz at timestamp 2
     add ref abc at timestamp 2
     => 00002.ref
  Transaction 2: add delete ref xyz at timestamp 3
     => 00003.ref

On compacting 00002 and 00003 together, you'll end up with a table for
only  "ref abc at timestamp 2". If you call that result 00002.ref
again, there are two versions 00002.ref.

In the current spec with unix semantics, that's not a problem, because
out of date tables are removed, and disappear once the last reader closes
the file, but if we have to cater for a system
where removal that doesn't always work, we'd have to fix up things
afterwards. That means we have to be able to tell apart these two
versions of 00002.ref, and this is where a random prefix would help.

The exact naming of the tables is not central to the spec, as the
table names to read are listed in tables.list, and systems can write
filenames as they please. Maybe recommendations around filenames
should be phrased as a "SHOULD" clause, with a "MUST" that each table
has a globally unique name.

> https://googlers.googlesource.com/sop/jgit/+/reftable/Documentation/technical/reftable.md#update-transactions
> https://googlers.googlesource.com/sop/jgit/+/reftable/Documentation/technical/reftable.md#header
>
> When the reftable format was initially being discussed I suggested
> somewhat flippantly (why not SQLite?). Partially because in trying to
> invent what's essentially a portable and concurrently updated database
> format we'd be likely to reinvent much of the same wheel.
>
> So not to drag that whole thing up again as a proposal for a format
> replacement, but I wonder what SQLite would do in this scenario & others
> where desired DB semantics / transactions and FS/portability semantics
> clash. AFAIK the reftable has only been battle-tested in production
> under JGit so far (presumably Linux-only). Whereas SQLite has been
> ported to and widely used on Windows, HP/UX and probably other systems
> where Linux-specific assumptions don't apply.

I think it would be an interesting exercise to build a ref database on
SQLite, and I wish it had been done before, because large part of the
work of getting reftable to work with git proper, is disentangling
the code with assumptions about the underlying ref storage system.

I agree that some of this is reinventing the wheel, and I wish that
reftable had been based on another format (say, LevelDB) to get away
from bit mangling, but I think SQlite is a tricky proposition, because
SQLite is a system with considerable complexity of its own. You'd
force all the other Git implementations (libgit2, Microsoft's C#,
go-git, JGit, Dulwhich etc.) to find a way to implement SQLite
support. That will be a difficult proposition for the implementations that
try to avoid FFI.

You could also look at this from the positive side, which is that the
transaction semantics of reftable are much easier to understand and
verify than those of the files backend.

--
Han-Wen Nienhuys - Google Munich
I work 80%. Don't expect answers from me on Fridays.
--

Google Germany GmbH, Erika-Mann-Strasse 33, 80636 Munich

Registergericht und -nummer: Hamburg, HRB 86891

Sitz der Gesellschaft: Hamburg

Geschäftsführer: Paul Manicle, Halimah DeLaine Prado
--

Google Germany GmbH, Erika-Mann-Strasse 33, 80636 Munich

Registergericht und -nummer: Hamburg, HRB 86891

Sitz der Gesellschaft: Hamburg

Geschäftsführer: Paul Manicle, Halimah DeLaine Prado