fix for dump headers

fix for dump headers

[David Bremner]

When I added message properties to the dump output I messed up the
copy-paste to add properties to the header. The fix is trivial, but
I'm not sure if this means we should bump the format version
(i.e. write #notmuch-dump:3) to denote the header can be relied
upon. I guess it doesn't hurt to do so. Opinions?

[PATCH 1/2] test: add known broken test for dump header

[David Bremner]

Apparently nobody uses the list of "what was included in the dump"
---
 test/T240-dump-restore.sh | 8 ++++++++
 1 file changed, 8 insertions(+)

diff --git a/test/T240-dump-restore.sh b/test/T240-dump-restore.sh
index 206c3ae1..f626a19e 100755
--- a/test/T240-dump-restore.sh
+++ b/test/T240-dump-restore.sh
@@ -2,6 +2,14 @@
 test_description="\"notmuch dump\" and \"notmuch restore\""
 . ./test-lib.sh || exit 1
 
+NOTMUCH_NEW > /dev/null
+test_begin_subtest "dump header"
+test_subtest_known_broken
+cat <<EOF > EXPECTED
+#notmuch-dump batch-tag:2 config,properties,tags
+EOF
+notmuch dump > OUTPUT
+test_expect_equal_file EXPECTED OUTPUT
 add_email_corpus
 
 test_begin_subtest "Dumping all tags"
-- 
2.11.0

Re: [PATCH 1/2] test: add known broken test for dump header

[David Bremner]

David Bremner <david@tethera.net> writes:

> Apparently nobody uses the list of "what was included in the dump"

test pushed to master.

[PATCH 2/2] cli/dump: fix bug in dump header

[David Bremner]

Fix copy paste error.
---
 notmuch-dump.c            | 2 +-
 test/T240-dump-restore.sh | 1 -
 2 files changed, 1 insertion(+), 2 deletions(-)

diff --git a/notmuch-dump.c b/notmuch-dump.c
index 0cbcdc16..e451ac0c 100644
--- a/notmuch-dump.c
+++ b/notmuch-dump.c
@@ -84,7 +84,7 @@ print_dump_header (gzFile output, int output_format, int include)
 	sep = ",";
     }
     if (include & DUMP_INCLUDE_TAGS) {
-	gzprintf (output, "%sproperties", sep);
+	gzprintf (output, "%stags", sep);
     }
     gzputs (output, "\n");
 }
diff --git a/test/T240-dump-restore.sh b/test/T240-dump-restore.sh
index f626a19e..a4eab65e 100755
--- a/test/T240-dump-restore.sh
+++ b/test/T240-dump-restore.sh
@@ -4,7 +4,6 @@ test_description="\"notmuch dump\" and \"notmuch restore\""
 
 NOTMUCH_NEW > /dev/null
 test_begin_subtest "dump header"
-test_subtest_known_broken
 cat <<EOF > EXPECTED
 #notmuch-dump batch-tag:2 config,properties,tags
 EOF
-- 
2.11.0

[PATCH] cli/dump: bump dump format version to 3

[David Bremner]

No changes to the format of the body, but the header format was
fixed, and version 2 headers probably shouldn't be relied on.
---
 doc/man1/notmuch-dump.rst | 2 +-
 notmuch-client.h          | 2 +-
 test/T240-dump-restore.sh | 2 +-
 test/T590-libconfig.sh    | 2 +-
 4 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/doc/man1/notmuch-dump.rst b/doc/man1/notmuch-dump.rst
index ca048aeb..f3f2b394 100644
--- a/doc/man1/notmuch-dump.rst
+++ b/doc/man1/notmuch-dump.rst
@@ -95,7 +95,7 @@ Supported options for **dump** include
 
       The default is to include all available types of data.  The
       option can be specified multiple times to select some subset. As
-      of version 2 of the dump format, there is a header line of the
+      of version 3 of the dump format, there is a header line of the
       following form
 
       |
diff --git a/notmuch-client.h b/notmuch-client.h
index 7a674873..a6f70eae 100644
--- a/notmuch-client.h
+++ b/notmuch-client.h
@@ -469,7 +469,7 @@ typedef enum dump_includes {
 
 #define DUMP_INCLUDE_DEFAULT (DUMP_INCLUDE_TAGS | DUMP_INCLUDE_CONFIG | DUMP_INCLUDE_PROPERTIES)
 
-#define NOTMUCH_DUMP_VERSION 2
+#define NOTMUCH_DUMP_VERSION 3
 
 int
 notmuch_database_dump (notmuch_database_t *notmuch,
diff --git a/test/T240-dump-restore.sh b/test/T240-dump-restore.sh
index a4eab65e..1f1f0866 100755
--- a/test/T240-dump-restore.sh
+++ b/test/T240-dump-restore.sh
@@ -5,7 +5,7 @@ test_description="\"notmuch dump\" and \"notmuch restore\""
 NOTMUCH_NEW > /dev/null
 test_begin_subtest "dump header"
 cat <<EOF > EXPECTED
-#notmuch-dump batch-tag:2 config,properties,tags
+#notmuch-dump batch-tag:3 config,properties,tags
 EOF
 notmuch dump > OUTPUT
 test_expect_equal_file EXPECTED OUTPUT
diff --git a/test/T590-libconfig.sh b/test/T590-libconfig.sh
index e8c078d5..1b308693 100755
--- a/test/T590-libconfig.sh
+++ b/test/T590-libconfig.sh
@@ -112,7 +112,7 @@ cat c_head - c_tail <<'EOF' | test_C ${MAIL_DIR}
 EOF
 notmuch dump --include=config >OUTPUT
 cat <<'EOF' >EXPECTED
-#notmuch-dump batch-tag:2 config
+#notmuch-dump batch-tag:3 config
 #@ aaabefore beforeval
 #@ key%20with%20spaces value,%20with,%20spaces%21
 #@ testkey1 testvalue1
-- 
2.11.0

Re: fix for dump headers

[Tomi Ollila]

On Tue, Mar 28 2017, David Bremner <david@tethera.net> wrote:

> When I added message properties to the dump output I messed up the
> copy-paste to add properties to the header. The fix is trivial, but
> I'm not sure if this means we should bump the format version
> (i.e. write #notmuch-dump:3) to denote the header can be relied
> upon. I guess it doesn't hurt to do so. Opinions?

I'd say bump the format version... so rest of the series LGTM.

Tomi

Re: fix for dump headers

[David Bremner]

Tomi Ollila <tomi.ollila@iki.fi> writes:

> On Tue, Mar 28 2017, David Bremner <david@tethera.net> wrote:
>
>> When I added message properties to the dump output I messed up the
>> copy-paste to add properties to the header. The fix is trivial, but
>> I'm not sure if this means we should bump the format version
>> (i.e. write #notmuch-dump:3) to denote the header can be relied
>> upon. I guess it doesn't hurt to do so. Opinions?
>
> I'd say bump the format version... so rest of the series LGTM.
>
> Tomi

OK, pushed to master and release

d