Skip to contentSkip to conversation

QError conversions and more

PATCH In review Markus Armbruster 5 participants 31 messages Dec 7, 2009 – Dec 8, 2009

lore
Share & tools
lore
#0 lore raw

cover QError conversions and more

This patch series converts monitor commands eject, change, closefd, getfd to QError.

There's a simple bonus fix in PATCH 3.

Finally, PATCH 18 puts a human-readable error description into the QMP error response. For background, see the lengthy thread ending at
http://lists.gnu.org/archive/html/qemu-devel/2009-11/msg01221.html

Series should be applied after Luiz's work.

Luiz Capitulino (2):
QError: new class for device encrypted errors
monitor: do_cont(): Don't ask for passwords

Markus Armbruster (16):
monitor: Fix double-prompt after "change vnc passwd BLA"
QError: Put error definitions in alphabetical order
QError: New QERR_DEVICE_LOCKED
QError: New QERR_DEVICE_NOT_REMOVABLE
monitor: convert do_eject() to QError
QError: New QERR_INVALID_BLOCK_FORMAT
QError: New QERR_SET_PASSWD_FAILED
QError: New QERR_VNC_SERVER_FAILED
monitor: convert do_change() to QObject, QError
QError: New QERR_FD_NOT_FOUND
monitor: convert do_closefd() to QError
QError: New QERR_FD_NOT_SUPPLIED
New QERR_INVALID_PARAMETER
QError: New QERR_TOO_MANY_FILES
monitor: convert do_getfd() to QError
QMP: add human-readable description to error response

 QMP/qmp-spec.txt |    5 ++-
 monitor.c        |   55 +++++++++++++++++++++++-----------
 qemu-monitor.hx  |    3 +-
 qerror.c         |   85 +++++++++++++++++++++++++++++++++++++++++++-----------
 qerror.h         |   56 ++++++++++++++++++++++++++++-------
 5 files changed, 155 insertions(+), 49 deletions(-)
#1 lore raw

PATCH 1/18 QError: new class for device encrypted errors

From: Luiz Capitulino <[email protected]>

---
 qerror.c |    4 ++++
 qerror.h |    3 +++
 2 files changed, 7 insertions(+), 0 deletions(-)
qerror.c +4 -0
index eb4ce33..d90529c 100644
@@ -45,6 +45,10 @@ static const QErrorStringTable qerror_table[] = {
         .desc        = "The command %(name) has not been found",
     },
     {
+        .error_fmt = QERR_DEVICE_ENCRYPTED,
+        .desc      = "The %(device) is encrypted",
+    },
+    {
         .error_fmt = QERR_DEVICE_NOT_FOUND,
         .desc      = "The %(device) device has not been found",
     },
qerror.h +3 -0
index 5198adf..c9fcf97 100644
@@ -41,6 +41,9 @@ QError *qobject_to_qerror(const QObject *obj);
 #define QERR_COMMAND_NOT_FOUND \
         "{ 'class': 'CommandNotFound', 'data': { 'name': %s } }"
 
+#define QERR_DEVICE_ENCRYPTED \
+        "{ 'class': 'DeviceEncrypted', 'data': { 'device': %s } }"
+
 #define QERR_DEVICE_NOT_FOUND \
         "{ 'class': 'DeviceNotFound', 'data': { 'device': %s } }"
 
Signature · 1 line
1.6.2.5
#2 lore raw

PATCH 2/18 monitor: do_cont(): Don't ask for passwords

In reply to Markus Armbruster

From: Luiz Capitulino <[email protected]>

The do_cont() function will ask the user to enter a password if a device is encrypted.

This is invalid under QMP, so we raise a QERR_DEVICE_ENCRYPTED error.

---
 monitor.c |   10 +++++++++-
 1 files changed, 9 insertions(+), 1 deletions(-)
monitor.c +9 -1
index 0ff208e..50c616d 100644
@@ -148,7 +148,10 @@ static void monitor_read_command(Monitor *mon, int show_prompt)
 static int monitor_read_password(Monitor *mon, ReadLineFunc *readline_func,
                                  void *opaque)
 {
-    if (mon->rs) {
+    if (monitor_ctrl_mode(mon)) {
+        qemu_error_new(QERR_MISSING_PARAMETER, "password");
+        return -EINVAL;
+    } else if (mon->rs) {
         readline_start(mon->rs, "Password: ", 1, readline_func, opaque);
         /* prompt is printed on return from the command handler */
         return 0;
@@ -4188,6 +4191,11 @@ void monitor_read_bdrv_key_start(Monitor *mon, BlockDriverState *bs,
         return;
     }
 
+    if (monitor_ctrl_mode(mon)) {
+        qemu_error_new(QERR_DEVICE_ENCRYPTED, bdrv_get_device_name(bs));
+        return;
+    }
+
     monitor_printf(mon, "%s (%s) is encrypted.\n", bdrv_get_device_name(bs),
                    bdrv_get_encrypted_filename(bs));
 
Signature · 1 line
1.6.2.5
#3 lore raw

PATCH 3/18 monitor: Fix double-prompt after "change vnc passwd BLA"

In reply to Markus Armbruster
---
 monitor.c |   11 ++++++++---
 1 files changed, 8 insertions(+), 3 deletions(-)
monitor.c +8 -3
index 50c616d..b9d3d92 100644
@@ -845,12 +845,17 @@ static void do_change_block(Monitor *mon, const char *device,
     monitor_read_bdrv_key_start(mon, bs, NULL, NULL);
 }
 
-static void change_vnc_password_cb(Monitor *mon, const char *password,
-                                   void *opaque)
+static void change_vnc_password(Monitor *mon, const char *password)
 {
     if (vnc_display_password(NULL, password) < 0)
         monitor_printf(mon, "could not set VNC server password\n");
 
+}
+
+static void change_vnc_password_cb(Monitor *mon, const char *password,
+                                   void *opaque)
+{
+    change_vnc_password(mon, password);
     monitor_read_command(mon, 1);
 }
 
@@ -862,7 +867,7 @@ static void do_change_vnc(Monitor *mon, const char *target, const char *arg)
             char password[9];
             strncpy(password, arg, sizeof(password));
             password[sizeof(password) - 1] = '\0';
-            change_vnc_password_cb(mon, password, NULL);
+            change_vnc_password(mon, password);
         } else {
             monitor_read_password(mon, change_vnc_password_cb, NULL);
         }
Signature · 1 line
1.6.2.5
#4 lore raw

PATCH 4/18 QError: Put error definitions in alphabetical order

In reply to Markus Armbruster

Also fix the odd typoe and clean up whitespace.

---
 qerror.c |   28 ++++++++++++++--------------
 qerror.h |   30 +++++++++++++++---------------
 2 files changed, 29 insertions(+), 29 deletions(-)
qerror.c +14 -14
index d90529c..03e3b34 100644
@@ -41,28 +41,32 @@ static const QType qerror_type = {
  */
 static const QErrorStringTable qerror_table[] = {
     {
-        .error_fmt   = QERR_COMMAND_NOT_FOUND,
-        .desc        = "The command %(name) has not been found",
+        .error_fmt = QERR_COMMAND_NOT_FOUND,
+        .desc      = "The command %(name) has not been found",
     },
     {
         .error_fmt = QERR_DEVICE_ENCRYPTED,
         .desc      = "The %(device) is encrypted",
     },
     {
+        .error_fmt = QERR_DEVICE_NOT_ACTIVE,
+        .desc      = "The %(device) device has not been activated by the guest",
+    },
+    {
         .error_fmt = QERR_DEVICE_NOT_FOUND,
         .desc      = "The %(device) device has not been found",
     },
     {
-        .error_fmt = QERR_DEVICE_NOT_ACTIVE,
-        .desc      = "The %(device) device has not been activated by the guest",
+        .error_fmt = QERR_INVALID_PARAMETER_TYPE,
+        .desc      = "Invalid parameter type, expected: %(expected)",
     },
     {
-        .error_fmt   = QERR_INVALID_PARAMETER_TYPE,
-        .desc        = "Invalid parameter type, expected: %(expected)",
+        .error_fmt = QERR_INVALID_PASSWORD,
+        .desc      = "The entered password is invalid",
     },
     {
-        .error_fmt   = QERR_INVALID_PASSWORD,
-        .desc        = "The entered password is invalid",
+        .error_fmt = QERR_JSON_PARSING,
+        .desc      = "Invalid JSON syntax",
     },
     {
         .error_fmt = QERR_KVM_MISSING_CAP,
@@ -77,12 +81,8 @@ static const QErrorStringTable qerror_table[] = {
         .desc      = "Bad QMP input object",
     },
     {
-        .error_fmt = QERR_JSON_PARSING,
-        .desc      = "Invalid JSON synaxt",
-    },
-    {
-        .error_fmt   = QERR_UNDEFINED_ERROR,
-        .desc        = "An undefined error has ocurred",
+        .error_fmt = QERR_UNDEFINED_ERROR,
+        .desc      = "An undefined error has ocurred",
     },
     {}
 };
qerror.h +15 -15
index c9fcf97..062c0c4 100644
@@ -39,36 +39,36 @@ QError *qobject_to_qerror(const QObject *obj);
  * QError class list
  */
 #define QERR_COMMAND_NOT_FOUND \
-        "{ 'class': 'CommandNotFound', 'data': { 'name': %s } }"
+    "{ 'class': 'CommandNotFound', 'data': { 'name': %s } }"
 
 #define QERR_DEVICE_ENCRYPTED \
-        "{ 'class': 'DeviceEncrypted', 'data': { 'device': %s } }"
-
-#define QERR_DEVICE_NOT_FOUND \
-        "{ 'class': 'DeviceNotFound', 'data': { 'device': %s } }"
+    "{ 'class': 'DeviceEncrypted', 'data': { 'device': %s } }"
 
 #define QERR_DEVICE_NOT_ACTIVE \
-        "{ 'class': 'DeviceNotActive', 'data': { 'device': %s } }"
+    "{ 'class': 'DeviceNotActive', 'data': { 'device': %s } }"
+
+#define QERR_DEVICE_NOT_FOUND \
+    "{ 'class': 'DeviceNotFound', 'data': { 'device': %s } }"
 
 #define QERR_INVALID_PARAMETER_TYPE \
-        "{ 'class': 'InvalidParameterType', 'data': { 'name': %s,'expected': %s } }"
+    "{ 'class': 'InvalidParameterType', 'data': { 'name': %s,'expected': %s } }"
 
 #define QERR_INVALID_PASSWORD \
-        "{ 'class': 'InvalidPassword', 'data': {} }"
+    "{ 'class': 'InvalidPassword', 'data': {} }"
+
+#define QERR_JSON_PARSING \
+    "{ 'class': 'JSONParsing', 'data': {} }"
 
 #define QERR_KVM_MISSING_CAP \
-        "{ 'class': 'KVMMissingCap', 'data': { 'capability': %s, 'feature': %s } }"
+    "{ 'class': 'KVMMissingCap', 'data': { 'capability': %s, 'feature': %s } }"
 
 #define QERR_MISSING_PARAMETER \
-        "{ 'class': 'MissingParameter', 'data': { 'name': %s } }"
+    "{ 'class': 'MissingParameter', 'data': { 'name': %s } }"
 
 #define QERR_QMP_BAD_INPUT_OBJECT \
-        "{ 'class': 'QMPBadInputObject', 'data': { 'expected': %s } }"
-
-#define QERR_JSON_PARSING \
-        "{ 'class': 'JSONParsing', 'data': {} }"
+    "{ 'class': 'QMPBadInputObject', 'data': { 'expected': %s } }"
 
 #define QERR_UNDEFINED_ERROR \
-        "{ 'class': 'UndefinedError', 'data': {} }"
+    "{ 'class': 'UndefinedError', 'data': {} }"
 
 #endif /* QERROR_H */
Signature · 1 line
1.6.2.5
#5 lore raw

PATCH 5/18 QError: New QERR_DEVICE_LOCKED

In reply to Markus Armbruster
---
 qerror.c |    4 ++++
 qerror.h |    3 +++
 2 files changed, 7 insertions(+), 0 deletions(-)
qerror.c +4 -0
index 03e3b34..37e8072 100644
@@ -49,6 +49,10 @@ static const QErrorStringTable qerror_table[] = {
         .desc      = "The %(device) is encrypted",
     },
     {
+        .error_fmt = QERR_DEVICE_LOCKED,
+        .desc      = "Device %(device) is locked",
+    },
+    {
         .error_fmt = QERR_DEVICE_NOT_ACTIVE,
         .desc      = "The %(device) device has not been activated by the guest",
     },
qerror.h +3 -0
index 062c0c4..be6cd68 100644
@@ -44,6 +44,9 @@ QError *qobject_to_qerror(const QObject *obj);
 #define QERR_DEVICE_ENCRYPTED \
     "{ 'class': 'DeviceEncrypted', 'data': { 'device': %s } }"
 
+#define QERR_DEVICE_LOCKED                                      \
+    "{ 'class': 'DeviceLocked', 'data': { 'device': %s } }"
+
 #define QERR_DEVICE_NOT_ACTIVE \
     "{ 'class': 'DeviceNotActive', 'data': { 'device': %s } }"
 
Signature · 1 line
1.6.2.5
#6 lore raw

PATCH 6/18 QError: New QERR_DEVICE_NOT_REMOVABLE

In reply to Markus Armbruster
---
 qerror.c |    4 ++++
 qerror.h |    3 +++
 2 files changed, 7 insertions(+), 0 deletions(-)
qerror.c +4 -0
index 37e8072..17532b0 100644
@@ -61,6 +61,10 @@ static const QErrorStringTable qerror_table[] = {
         .desc      = "The %(device) device has not been found",
     },
     {
+        .error_fmt = QERR_DEVICE_NOT_REMOVABLE,
+        .desc      = "Device %(device) is not removable",
+    },
+    {
         .error_fmt = QERR_INVALID_PARAMETER_TYPE,
         .desc      = "Invalid parameter type, expected: %(expected)",
     },
qerror.h +3 -0
index be6cd68..2abff1e 100644
@@ -53,6 +53,9 @@ QError *qobject_to_qerror(const QObject *obj);
 #define QERR_DEVICE_NOT_FOUND \
     "{ 'class': 'DeviceNotFound', 'data': { 'device': %s } }"
 
+#define QERR_DEVICE_NOT_REMOVABLE \
+    "{ 'class': 'DeviceNotRemovable', 'data': { 'device': %s } }"
+
 #define QERR_INVALID_PARAMETER_TYPE \
     "{ 'class': 'InvalidParameterType', 'data': { 'name': %s,'expected': %s } }"
 
Signature · 1 line
1.6.2.5
#7 lore raw

PATCH 7/18 monitor: convert do_eject() to QError

In reply to Markus Armbruster

Also affects do_change(), because the two share eject_device().

---
 monitor.c |    7 ++++---
 1 files changed, 4 insertions(+), 3 deletions(-)
monitor.c +4 -3
index b9d3d92..62d2ef5 100644
@@ -778,11 +778,12 @@ static int eject_device(Monitor *mon, BlockDriverState *bs, int force)
     if (bdrv_is_inserted(bs)) {
         if (!force) {
             if (!bdrv_is_removable(bs)) {
-                monitor_printf(mon, "device is not removable\n");
+                qemu_error_new(QERR_DEVICE_NOT_REMOVABLE,
+                               bdrv_get_device_name(bs));
                 return -1;
             }
             if (bdrv_is_locked(bs)) {
-                monitor_printf(mon, "device is locked\n");
+                qemu_error_new(QERR_DEVICE_LOCKED, bdrv_get_device_name(bs));
                 return -1;
             }
         }
@@ -799,7 +800,7 @@ static void do_eject(Monitor *mon, const QDict *qdict, QObject **ret_data)
 
     bs = bdrv_find(filename);
     if (!bs) {
-        monitor_printf(mon, "device not found\n");
+        qemu_error_new(QERR_DEVICE_NOT_FOUND, filename);
         return;
     }
     eject_device(mon, bs, force);
Signature · 1 line
1.6.2.5
#8 lore raw

PATCH 8/18 QError: New QERR_INVALID_BLOCK_FORMAT

In reply to Markus Armbruster
---
 qerror.c |    4 ++++
 qerror.h |    3 +++
 2 files changed, 7 insertions(+), 0 deletions(-)
qerror.c +4 -0
index 17532b0..56083c4 100644
@@ -65,6 +65,10 @@ static const QErrorStringTable qerror_table[] = {
         .desc      = "Device %(device) is not removable",
     },
     {
+        .error_fmt = QERR_INVALID_BLOCK_FORMAT,
+        .desc      = "Invalid block format %(name)",
+    },
+    {
         .error_fmt = QERR_INVALID_PARAMETER_TYPE,
         .desc      = "Invalid parameter type, expected: %(expected)",
     },
qerror.h +3 -0
index 2abff1e..681390c 100644
@@ -56,6 +56,9 @@ QError *qobject_to_qerror(const QObject *obj);
 #define QERR_DEVICE_NOT_REMOVABLE \
     "{ 'class': 'DeviceNotRemovable', 'data': { 'device': %s } }"
 
+#define QERR_INVALID_BLOCK_FORMAT \
+    "{ 'class': 'InvalidBlockFormat', 'data': { 'name': %s } }"
+
 #define QERR_INVALID_PARAMETER_TYPE \
     "{ 'class': 'InvalidParameterType', 'data': { 'name': %s,'expected': %s } }"
 
Signature · 1 line
1.6.2.5
#9 lore raw

PATCH 9/18 QError: New QERR_SET_PASSWD_FAILED

In reply to Markus Armbruster
---
 qerror.c |    4 ++++
 qerror.h |    3 +++
 2 files changed, 7 insertions(+), 0 deletions(-)
qerror.c +4 -0
index 56083c4..1bb68c4 100644
@@ -93,6 +93,10 @@ static const QErrorStringTable qerror_table[] = {
         .desc      = "Bad QMP input object",
     },
     {
+        .error_fmt = QERR_SET_PASSWD_FAILED,
+        .desc      = "Could not set password",
+    },
+    {
         .error_fmt = QERR_UNDEFINED_ERROR,
         .desc      = "An undefined error has ocurred",
     },
qerror.h +3 -0
index 681390c..b856899 100644
@@ -77,6 +77,9 @@ QError *qobject_to_qerror(const QObject *obj);
 #define QERR_QMP_BAD_INPUT_OBJECT \
     "{ 'class': 'QMPBadInputObject', 'data': { 'expected': %s } }"
 
+#define QERR_SET_PASSWD_FAILED \
+    "{ 'class': 'SetPasswdFailed', 'data': {} }"
+
 #define QERR_UNDEFINED_ERROR \
     "{ 'class': 'UndefinedError', 'data': {} }"
 
Signature · 1 line
1.6.2.5
#10 lore raw

PATCH 10/18 QError: New QERR_VNC_SERVER_FAILED

In reply to Markus Armbruster
---
 qerror.c |    4 ++++
 qerror.h |    3 +++
 2 files changed, 7 insertions(+), 0 deletions(-)
qerror.c +4 -0
index 1bb68c4..128a91e 100644
@@ -100,6 +100,10 @@ static const QErrorStringTable qerror_table[] = {
         .error_fmt = QERR_UNDEFINED_ERROR,
         .desc      = "An undefined error has ocurred",
     },
+    {
+        .error_fmt = QERR_VNC_SERVER_FAILED,
+        .desc      = "Could not start VNC server on %(target)",
+    },
     {}
 };
 
qerror.h +3 -0
index b856899..1a05d9a 100644
@@ -83,4 +83,7 @@ QError *qobject_to_qerror(const QObject *obj);
 #define QERR_UNDEFINED_ERROR \
     "{ 'class': 'UndefinedError', 'data': {} }"
 
+#define QERR_VNC_SERVER_FAILED \
+    "{ 'class': 'VNCServerFailed', 'data': { 'target': %s } }"
+
 #endif /* QERROR_H */
Signature · 1 line
1.6.2.5
#11 lore raw

PATCH 11/18 monitor: convert do_change() to QObject, QError

In reply to Markus Armbruster
---
 monitor.c       |   19 +++++++++++--------
 qemu-monitor.hx |    3 ++-
 2 files changed, 13 insertions(+), 9 deletions(-)
monitor.c +11 -8
index 62d2ef5..05233cc 100644
@@ -830,13 +830,13 @@ static void do_change_block(Monitor *mon, const char *device,
 
     bs = bdrv_find(device);
     if (!bs) {
-        monitor_printf(mon, "device not found\n");
+        qemu_error_new(QERR_DEVICE_NOT_FOUND, device);
         return;
     }
     if (fmt) {
         drv = bdrv_find_whitelisted_format(fmt);
         if (!drv) {
-            monitor_printf(mon, "invalid format %s\n", fmt);
+            qemu_error_new(QERR_INVALID_BLOCK_FORMAT, fmt);
             return;
         }
     }
@@ -846,17 +846,17 @@ static void do_change_block(Monitor *mon, const char *device,
     monitor_read_bdrv_key_start(mon, bs, NULL, NULL);
 }
 
-static void change_vnc_password(Monitor *mon, const char *password)
+static void change_vnc_password(const char *password)
 {
     if (vnc_display_password(NULL, password) < 0)
-        monitor_printf(mon, "could not set VNC server password\n");
+        qemu_error_new(QERR_SET_PASSWD_FAILED);
 
 }
 
 static void change_vnc_password_cb(Monitor *mon, const char *password,
                                    void *opaque)
 {
-    change_vnc_password(mon, password);
+    change_vnc_password(password);
     monitor_read_command(mon, 1);
 }
 
@@ -868,17 +868,20 @@ static void do_change_vnc(Monitor *mon, const char *target, const char *arg)
             char password[9];
             strncpy(password, arg, sizeof(password));
             password[sizeof(password) - 1] = '\0';
-            change_vnc_password(mon, password);
+            change_vnc_password(password);
         } else {
             monitor_read_password(mon, change_vnc_password_cb, NULL);
         }
     } else {
         if (vnc_display_open(NULL, target) < 0)
-            monitor_printf(mon, "could not start VNC server on %s\n", target);
+            qemu_error_new(QERR_VNC_SERVER_FAILED, target);
     }
 }
 
-static void do_change(Monitor *mon, const QDict *qdict)
+/**
+ * do_change(): Change a removable medium, or VNC configuration
+ */
+static void do_change(Monitor *mon, const QDict *qdict, QObject **ret_data)
 {
     const char *device = qdict_get_str(qdict, "device");
     const char *target = qdict_get_str(qdict, "target");
qemu-monitor.hx +2 -1
index cf2ef9b..c788c73 100644
@@ -147,7 +147,8 @@ ETEXI
         .args_type  = "device:B,target:F,arg:s?",
         .params     = "device filename [format]",
         .help       = "change a removable medium, optional format",
-        .mhandler.cmd = do_change,
+        .user_print = monitor_user_noop,
+        .mhandler.cmd_new = do_change,
     },
 
 STEXI
Signature · 1 line
1.6.2.5
#12 lore raw

PATCH 12/18 QError: New QERR_FD_NOT_FOUND

In reply to Markus Armbruster
---
 qerror.c |    4 ++++
 qerror.h |    3 +++
 2 files changed, 7 insertions(+), 0 deletions(-)
qerror.c +4 -0
index 128a91e..d75c69a 100644
@@ -65,6 +65,10 @@ static const QErrorStringTable qerror_table[] = {
         .desc      = "Device %(device) is not removable",
     },
     {
+        .error_fmt = QERR_FD_NOT_FOUND,
+        .desc      = "Failed to find file descriptor named %(name)",
+    },
+    {
         .error_fmt = QERR_INVALID_BLOCK_FORMAT,
         .desc      = "Invalid block format %(name)",
     },
qerror.h +3 -0
index 1a05d9a..bda764f 100644
@@ -56,6 +56,9 @@ QError *qobject_to_qerror(const QObject *obj);
 #define QERR_DEVICE_NOT_REMOVABLE \
     "{ 'class': 'DeviceNotRemovable', 'data': { 'device': %s } }"
 
+#define QERR_FD_NOT_FOUND \
+    "{ 'class': 'fd_not_found', 'data': { 'name': %s } }"
+
 #define QERR_INVALID_BLOCK_FORMAT \
     "{ 'class': 'InvalidBlockFormat', 'data': { 'name': %s } }"
 
Signature · 1 line
1.6.2.5
#13 lore raw

[Qemu-devel] Re: [FOR 0.12 PATCH 12/18] QError: New QERR_FD_NOT_FOUND

qerror.h QError *qobject_to_qerror(const QObject *obj); · line 61

Quoted 3 lines · On Mon, 7 Dec 2009 21:37:10 +0100 Markus Armbruster <[email protected]> wrote:
qerror.h +3 -0
+#define QERR_FD_NOT_FOUND \
+    "{ 'class': 'fd_not_found', 'data': { 'name': %s } }"
+

I think this came from your previous series of not using CamelCase?

Also happens in next patches..

#14 lore raw

[Qemu-devel] Re: [FOR 0.12 PATCH 12/18] QError: New QERR_FD_NOT_FOUND

Quoted 10 lines · Luiz Capitulino

qerror.h QError *qobject_to_qerror(const QObject *obj); · line 61

Quoted 3 lines · On Mon, 7 Dec 2009 21:37:10 +0100 Markus Armbruster <[email protected]> wrote:
qerror.h +3 -0
+#define QERR_FD_NOT_FOUND \
+    "{ 'class': 'fd_not_found', 'data': { 'name': %s } }"
+

I think this came from your previous series of not using CamelCase?

Also happens in next patches..

Rats. Fixing...

#15 lore raw

PATCH 13/18 monitor: convert do_closefd() to QError

In reply to Markus Armbruster
---
 monitor.c |    3 +--
 1 files changed, 1 insertions(+), 2 deletions(-)
monitor.c +1 -2
index 05233cc..c35a31e 100644
@@ -2195,8 +2195,7 @@ static void do_closefd(Monitor *mon, const QDict *qdict, QObject **ret_data)
         return;
     }
 
-    monitor_printf(mon, "Failed to find file descriptor named %s\n",
-                   fdname);
+    qemu_error_new(QERR_FD_NOT_FOUND, fdname);
 }
 
 static void do_loadvm(Monitor *mon, const QDict *qdict)
Signature · 1 line
1.6.2.5
#16 lore raw

PATCH 14/18 QError: New QERR_FD_NOT_SUPPLIED

In reply to Markus Armbruster
---
 qerror.c |    4 ++++
 qerror.h |    3 +++
 2 files changed, 7 insertions(+), 0 deletions(-)
qerror.c +4 -0
index d75c69a..e6b7f62 100644
@@ -69,6 +69,10 @@ static const QErrorStringTable qerror_table[] = {
         .desc      = "Failed to find file descriptor named %(name)",
     },
     {
+        .error_fmt = QERR_FD_NOT_SUPPLIED,
+        .desc      = "No file descriptor supplied via SCM_RIGHTS",
+    },
+    {
         .error_fmt = QERR_INVALID_BLOCK_FORMAT,
         .desc      = "Invalid block format %(name)",
     },
qerror.h +3 -0
index bda764f..2262619 100644
@@ -59,6 +59,9 @@ QError *qobject_to_qerror(const QObject *obj);
 #define QERR_FD_NOT_FOUND \
     "{ 'class': 'fd_not_found', 'data': { 'name': %s } }"
 
+#define QERR_FD_NOT_SUPPLIED \
+    "{ 'class': 'fd_not_supplied', 'data': {} }"
+
 #define QERR_INVALID_BLOCK_FORMAT \
     "{ 'class': 'InvalidBlockFormat', 'data': { 'name': %s } }"
 
Signature · 1 line
1.6.2.5
#17 lore raw

PATCH 15/18 New QERR_INVALID_PARAMETER

In reply to Markus Armbruster
---
 qerror.c |    4 ++++
 qerror.h |    3 +++
 2 files changed, 7 insertions(+), 0 deletions(-)
qerror.c +4 -0
index e6b7f62..aa89a3d 100644
@@ -77,6 +77,10 @@ static const QErrorStringTable qerror_table[] = {
         .desc      = "Invalid block format %(name)",
     },
     {
+        .error_fmt = QERR_INVALID_PARAMETER,
+        .desc      = "Invalid parameter %(name)",
+    },
+    {
         .error_fmt = QERR_INVALID_PARAMETER_TYPE,
         .desc      = "Invalid parameter type, expected: %(expected)",
     },
qerror.h +3 -0
index 2262619..48f1ab6 100644
@@ -65,6 +65,9 @@ QError *qobject_to_qerror(const QObject *obj);
 #define QERR_INVALID_BLOCK_FORMAT \
     "{ 'class': 'InvalidBlockFormat', 'data': { 'name': %s } }"
 
+#define QERR_INVALID_PARAMETER \
+    "{ 'class': 'invalid_parameter', 'data': { 'name': %s } }"
+
 #define QERR_INVALID_PARAMETER_TYPE \
     "{ 'class': 'InvalidParameterType', 'data': { 'name': %s,'expected': %s } }"
 
Signature · 1 line
1.6.2.5
#18 lore raw

PATCH 16/18 QError: New QERR_TOO_MANY_FILES

In reply to Markus Armbruster
---
 qerror.c |    4 ++++
 qerror.h |    3 +++
 2 files changed, 7 insertions(+), 0 deletions(-)
qerror.c +4 -0
index aa89a3d..8ffe4f6 100644
@@ -109,6 +109,10 @@ static const QErrorStringTable qerror_table[] = {
         .desc      = "Could not set password",
     },
     {
+        .error_fmt = QERR_TOO_MANY_FILES,
+        .desc      = "Too many open files",
+    },
+    {
         .error_fmt = QERR_UNDEFINED_ERROR,
         .desc      = "An undefined error has ocurred",
     },
qerror.h +3 -0
index 48f1ab6..9462d5c 100644
@@ -92,6 +92,9 @@ QError *qobject_to_qerror(const QObject *obj);
 #define QERR_UNDEFINED_ERROR \
     "{ 'class': 'UndefinedError', 'data': {} }"
 
+#define QERR_TOO_MANY_FILES \
+    "{ 'class': 'fd_too_many_files', 'data': {} }"
+
 #define QERR_VNC_SERVER_FAILED \
     "{ 'class': 'VNCServerFailed', 'data': { 'target': %s } }"
 
Signature · 1 line
1.6.2.5
#19 lore raw

PATCH 17/18 monitor: convert do_getfd() to QError

In reply to Markus Armbruster
---
 monitor.c |   10 ++++++----
 1 files changed, 6 insertions(+), 4 deletions(-)
monitor.c +6 -4
index c35a31e..0bcffbe 100644
@@ -2145,19 +2145,21 @@ static void do_getfd(Monitor *mon, const QDict *qdict, QObject **ret_data)
 
     fd = qemu_chr_get_msgfd(mon->chr);
     if (fd == -1) {
-        monitor_printf(mon, "getfd: no file descriptor supplied via SCM_RIGHTS\n");
+        qemu_error_new(QERR_FD_NOT_SUPPLIED);
         return;
     }
 
     if (qemu_isdigit(fdname[0])) {
-        monitor_printf(mon, "getfd: monitor names may not begin with a number\n");
+        qemu_error_new(QERR_INVALID_PARAMETER, "fdname");
         return;
     }
 
     fd = dup(fd);
     if (fd == -1) {
-        monitor_printf(mon, "Failed to dup() file descriptor: %s\n",
-                       strerror(errno));
+        if (errno == EMFILE)
+            qemu_error_new(QERR_TOO_MANY_FILES);
+        else
+            qemu_error_new(QERR_UNDEFINED_ERROR);
         return;
     }
 
Signature · 1 line
1.6.2.5
#20 lore raw

PATCH 18/18 QMP: add human-readable description to error response

In reply to Markus Armbruster
---
 QMP/qmp-spec.txt |    5 ++++-
 monitor.c        |    1 +
 qerror.c         |   21 ++++++++++++++++-----
 qerror.h         |    2 ++
 4 files changed, 23 insertions(+), 6 deletions(-)
QMP/qmp-spec.txt +4 -1
index 8429789..1cbd21c 100644
@@ -102,13 +102,16 @@ completed because of an error condition.
 
 The format is:
 
-{ "error": { "class": json-string, "data": json-value }, "id": json-value }
+{ "error": { "class": json-string, "data": json-value, "desc": json-string },
+  "id": json-value }
 
  Where,
 
 - The "class" member contains the error class name (eg. "ServiceUnavailable")
 - The "data" member contains specific error data and is defined in a
   per-command basis, it will be an empty json-object if the error has no data
+- The "desc" member is a human-readable error message.  Clients should
+  not attempt to parse this message.
 - The "id" member contains the transaction identification associated with
   the command execution (if issued by the Client)
 
monitor.c +1 -0
index 0bcffbe..4ad1b5e 100644
@@ -305,6 +305,7 @@ static void monitor_protocol_emitter(Monitor *mon, QObject *data)
         }
     } else {
         /* error response */
+        qdict_put(mon->error->error, "desc", qerror_human(mon->error));
         qdict_put(qmp, "error", mon->error->error);
         QINCREF(mon->error->error);
         QDECREF(mon->error);
qerror.c +16 -5
index 8ffe4f6..5f8fc5d 100644
@@ -283,13 +283,11 @@ static const char *append_field(QString *outstr, const QError *qerror,
 }
 
 /**
- * qerror_print(): Print QError data
+ * qerror_human(): Format QError data into human-readable string.
  *
- * This function will print the member 'desc' of the specified QError object,
- * it uses qemu_error() for this, so that the output is routed to the right
- * place (ie. stderr or Monitor's device).
+ * Formats according to member 'desc' of the specified QError object.
  */
-void qerror_print(const QError *qerror)
+QString *qerror_human(const QError *qerror)
 {
     const char *p;
     QString *qstring;
@@ -309,6 +307,19 @@ void qerror_print(const QError *qerror)
         }
     }
 
+    return qstring;
+}
+
+/**
+ * qerror_print(): Print QError data
+ *
+ * This function will print the member 'desc' of the specified QError object,
+ * it uses qemu_error() for this, so that the output is routed to the right
+ * place (ie. stderr or Monitor's device).
+ */
+void qerror_print(const QError *qerror)
+{
+    QString *qstring = qerror_human(qerror);
     qemu_error("%s\n", qstring_get_str(qstring));
     QDECREF(qstring);
 }
qerror.h +2 -0
index 9462d5c..09e32b9 100644
@@ -13,6 +13,7 @@
 #define QERROR_H
 
 #include "qdict.h"
+#include "qstring.h"
 #include <stdarg.h>
 
 typedef struct QErrorStringTable {
@@ -32,6 +33,7 @@ typedef struct QError {
 QError *qerror_new(void);
 QError *qerror_from_info(const char *file, int linenr, const char *func,
                          const char *fmt, va_list *va);
+QString *qerror_human(const QError *qerror);
 void qerror_print(const QError *qerror);
 QError *qobject_to_qerror(const QObject *obj);
 
Signature · 1 line
1.6.2.5
#21 lore raw

[Qemu-devel] Re: [FOR 0.12 PATCH 18/18] QMP: add human-readable description to error response

QMP/qmp-spec.txt completed because of an error condition. · line 116

Quoted 13 lines · On Mon, 7 Dec 2009 21:37:16 +0100 Markus Armbruster <[email protected]> wrote:
QMP/qmp-spec.txt +4 -1
-{ "error": { "class": json-string, "data": json-value }, "id": json-value }
+{ "error": { "class": json-string, "data": json-value, "desc": json-string },
+  "id": json-value }
 
  Where,
 
 - The "class" member contains the error class name (eg. "ServiceUnavailable")
 - The "data" member contains specific error data and is defined in a
   per-command basis, it will be an empty json-object if the error has no data
+- The "desc" member is a human-readable error message.  Clients should
+  not attempt to parse this message.
 - The "id" member contains the transaction identification associated with
   the command execution (if issued by the Client)

As we've talked on irc, I don't agree with this change.

Basically, adding 'desc' to the standard error message introduces all
the problems we've discussed about free-form English strings.

I feel that QError is becoming the worst of all proposals.

I agree with you that it's not as easy as it should be to report errors,
but as we're targeting on Clients I was convinced that we could not have the best API internally but offer a good interface for Clients.

Now, having 'desc' as part of the standard protocol is like not having
the best API internally and offering a bad interface for Clients.

Not to mention that those strings can't be modified when the protocol
becomes stable and we're probably talking about dozens if not a hundred of strings. Ok, there isn't a reason to change them often, but it's still one more thing to maintain.

Having said that, I would agree to have 'desc' as part of debug
information. I have patches in my tree which adds CONFIG_DEBUG_QMP, if one enables it information about the error location will also be part of the error message. I would agree having 'desc' there too.

#22 lore raw

Re: [Qemu-devel] Re: [FOR 0.12 PATCH 18/18] QMP: add human-readable description to error response

Quoted 28 lines · Luiz Capitulino

QMP/qmp-spec.txt completed because of an error condition. · line 116

Quoted 13 lines · On Mon, 7 Dec 2009 21:37:16 +0100 Markus Armbruster <[email protected]> wrote:
QMP/qmp-spec.txt +4 -1
-{ "error": { "class": json-string, "data": json-value }, "id": json-value }
+{ "error": { "class": json-string, "data": json-value, "desc": json-string },
+  "id": json-value }
 
  Where,
 
 - The "class" member contains the error class name (eg. "ServiceUnavailable")
 - The "data" member contains specific error data and is defined in a
   per-command basis, it will be an empty json-object if the error has no data
+- The "desc" member is a human-readable error message.  Clients should
+  not attempt to parse this message.
 - The "id" member contains the transaction identification associated with
   the command execution (if issued by the Client)

As we've talked on irc, I don't agree with this change.

Basically, adding 'desc' to the standard error message introduces all
the problems we've discussed about free-form English strings.

I feel that QError is becoming the worst of all proposals.

I agree with you that it's not as easy as it should be to report errors,
but as we're targeting on Clients I was convinced that we could not have the best API internally but offer a good interface for Clients.

Quoted 7 lines

Now, having 'desc' as part of the standard protocol is like not having
the best API internally and offering a bad interface for Clients.

Not to mention that those strings can't be modified when the protocol
becomes stable and we're probably talking about dozens if not a hundred of strings. Ok, there isn't a reason to change them often, but it's still one more thing to maintain.

I think it is fine to declare that 'desc' strings are subject to arbitrary change. Even if QEMU includes this 'desc' I think in libvirt will end up doing its own error code -> human string conversion, simply because we need to translate the strings. So we'd only use 'desc' in any logging calls if it were present.

Quoted 5 lines

Having said that, I would agree to have 'desc' as part of debug
information. I have patches in my tree which adds CONFIG_DEBUG_QMP, if one enables it information about the error location will also be part of the error message. I would agree having 'desc' there too.

Daniel

Signature · 4 lines
|: Red Hat, Engineering, London   -o-   http://people.redhat.com/berrange/ :|
|: http://libvirt.org  -o-  http://virt-manager.org  -o-  http://ovirt.org :|
|: http://autobuild.org       -o-         http://search.cpan.org/~danberr/ :|
|: GnuPG: 7D3B9505  -o-  F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 :|
#23 lore raw

Re: [Qemu-devel] Re: [FOR 0.12 PATCH 18/18] QMP: add human-readable description to error response

Quoted 37 lines · On Tue, 8 Dec 2009 12:25:13 +0000 "Daniel P. Berrange" <[email protected]>…
Quoted 28 lines · Luiz Capitulino

QMP/qmp-spec.txt completed because of an error condition. · line 116

Quoted 13 lines · On Mon, 7 Dec 2009 21:37:16 +0100 Markus Armbruster <[email protected]> wrote:
QMP/qmp-spec.txt +4 -1
-{ "error": { "class": json-string, "data": json-value }, "id": json-value }
+{ "error": { "class": json-string, "data": json-value, "desc": json-string },
+  "id": json-value }
 
  Where,
 
 - The "class" member contains the error class name (eg. "ServiceUnavailable")
 - The "data" member contains specific error data and is defined in a
   per-command basis, it will be an empty json-object if the error has no data
+- The "desc" member is a human-readable error message.  Clients should
+  not attempt to parse this message.
 - The "id" member contains the transaction identification associated with
   the command execution (if issued by the Client)

As we've talked on irc, I don't agree with this change.

Basically, adding 'desc' to the standard error message introduces all
the problems we've discussed about free-form English strings.

I feel that QError is becoming the worst of all proposals.

I agree with you that it's not as easy as it should be to report errors,
but as we're targeting on Clients I was convinced that we could not have the best API internally but offer a good interface for Clients.

Quoted 7 lines

Now, having 'desc' as part of the standard protocol is like not having
the best API internally and offering a bad interface for Clients.

Not to mention that those strings can't be modified when the protocol
becomes stable and we're probably talking about dozens if not a hundred of strings. Ok, there isn't a reason to change them often, but it's still one more thing to maintain.

Quoted 5 lines

I think it is fine to declare that 'desc' strings are subject to arbitrary change. Even if QEMU includes this 'desc' I think in libvirt will end up doing its own error code -> human string conversion, simply because we need to translate the strings. So we'd only use 'desc' in any logging calls if it were present.

That's what we would expect from clients, but I'm not convinced that
they'll took the easy way and start using it the way they shouldn't.

#24 lore raw

Re: [Qemu-devel] Re: [FOR 0.12 PATCH 18/18] QMP: add human-readable description to error response

Quoted 40 lines · Luiz Capitulino
Quoted 37 lines · On Tue, 8 Dec 2009 12:25:13 +0000 "Daniel P. Berrange" <[email protected]>…
Quoted 35 lines · Luiz Capitulino

QMP/qmp-spec.txt completed because of an error condition. · line 116

Quoted 13 lines · On Mon, 7 Dec 2009 21:37:16 +0100 Markus Armbruster <[email protected]> wrote:
QMP/qmp-spec.txt +4 -1
-{ "error": { "class": json-string, "data": json-value }, "id": json-value }
+{ "error": { "class": json-string, "data": json-value, "desc": json-string },
+  "id": json-value }
 
  Where,
 
 - The "class" member contains the error class name (eg. "ServiceUnavailable")
 - The "data" member contains specific error data and is defined in a
   per-command basis, it will be an empty json-object if the error has no data
+- The "desc" member is a human-readable error message.  Clients should
+  not attempt to parse this message.
 - The "id" member contains the transaction identification associated with
   the command execution (if issued by the Client)

As we've talked on irc, I don't agree with this change.

Basically, adding 'desc' to the standard error message introduces all
the problems we've discussed about free-form English strings.

I feel that QError is becoming the worst of all proposals.

I agree with you that it's not as easy as it should be to report errors,
but as we're targeting on Clients I was convinced that we could not have the best API internally but offer a good interface for Clients.

Now, having 'desc' as part of the standard protocol is like not having
the best API internally and offering a bad interface for Clients.

Not to mention that those strings can't be modified when the protocol
becomes stable and we're probably talking about dozens if not a hundred of strings. Ok, there isn't a reason to change them often, but it's still one more thing to maintain.

Quoted 8 lines
Quoted 5 lines

I think it is fine to declare that 'desc' strings are subject to arbitrary change. Even if QEMU includes this 'desc' I think in libvirt will end up doing its own error code -> human string conversion, simply because we need to translate the strings. So we'd only use 'desc' in any logging calls if it were present.

That's what we would expect from clients, but I'm not convinced that
they'll took the easy way and start using it the way they shouldn't.

Change a couple of common messages in every release, and they'll stop doing that real fast ;)

#25 lore raw

[Qemu-devel] Re: [FOR 0.12 PATCH 18/18] QMP: add human-readable description to error response

In reply to Luiz Capitulino
Quoted 3 lines · Luiz Capitulino

Not to mention that those strings can't be modified when the protocol
becomes stable and we're probably talking about dozens if not a hundred of strings.

I would say that there is _explicitly no promise_ of keeping these stable. You can serve

HTTP/1.1 302 You have bad taste in music
Location: http://www.britneyspears.com/

and clients will not complain. It's the same for QMP and desc.

(Disclaimer: I haven't read the entire series from Markus).

Paolo

#26 lore raw

[Qemu-devel] Re: [FOR 0.12 PATCH 18/18] QMP: add human-readable description to error response

Quoted 5 lines · Paolo Bonzini
Quoted 3 lines · Luiz Capitulino

Not to mention that those strings can't be modified when the protocol
becomes stable and we're probably talking about dozens if not a hundred of strings.

Quoted 7 lines

I would say that there is _explicitly no promise_ of keeping these stable. You can serve

HTTP/1.1 302 You have bad taste in music
Location: http://www.britneyspears.com/

and clients will not complain.

And even serve later in the same session

HTTP/1.1 302 Bad taste in music you have
Location: http://www.britneyspears.com/

Quoted 1 line

It's the same for QMP and desc.

Precisely.

Quoted 1 line

(Disclaimer: I haven't read the entire series from Markus).

#27 lore raw

[Qemu-devel] Re: [FOR 0.12 PATCH 18/18] QMP: add human-readable description to error response

In reply to Luiz Capitulino
Quoted 19 lines · Luiz Capitulino

QMP/qmp-spec.txt completed because of an error condition. · line 116

Quoted 13 lines · On Mon, 7 Dec 2009 21:37:16 +0100 Markus Armbruster <[email protected]> wrote:
QMP/qmp-spec.txt +4 -1
-{ "error": { "class": json-string, "data": json-value }, "id": json-value }
+{ "error": { "class": json-string, "data": json-value, "desc": json-string },
+  "id": json-value }
 
  Where,
 
 - The "class" member contains the error class name (eg. "ServiceUnavailable")
 - The "data" member contains specific error data and is defined in a
   per-command basis, it will be an empty json-object if the error has no data
+- The "desc" member is a human-readable error message.  Clients should
+  not attempt to parse this message.
 - The "id" member contains the transaction identification associated with
   the command execution (if issued by the Client)

As we've talked on irc, I don't agree with this change.

Quoted 2 lines

Basically, adding 'desc' to the standard error message introduces all
the problems we've discussed about free-form English strings.

I disagree. See below.

Quoted 1 line

I feel that QError is becoming the worst of all proposals.

I'm not exactly thrilled about it either, but it could clearly be worse in many ways, so it can't be the worst of all :)

Quoted 9 lines

I agree with you that it's not as easy as it should be to report errors,
but as we're targeting on Clients I was convinced that we could not have the best API internally but offer a good interface for Clients.

Now, having 'desc' as part of the standard protocol is like not having
the best API internally and offering a bad interface for Clients.

Not to mention that those strings can't be modified when the protocol
becomes stable

This is incorrect. We explicitly threaten client writers that these strings are not to be interpreted, in qmp-spec.txt: "Clients should not attempt to parse this message." For me, that implies that they can change at any time. Maybe we could use even stronger language there.

Quoted 9 lines

and we're probably talking about dozens if not a hundred
of strings. Ok, there isn't a reason to change them often, but it's still one more thing to maintain.

Having said that, I would agree to have 'desc' as part of debug
information. I have patches in my tree which adds CONFIG_DEBUG_QMP, if one enables it information about the error location will also be part of the error message. I would agree having 'desc' there too.

Debugging is one use for human-readable text in the error reply. But it is *not* the only use. We discussed this at length in October[*], and again in November[**]. Any new arguments?

[*] Sub-thread starting at
http://lists.gnu.org/archive/html/qemu-devel/2009-10/msg01245.html

[**] Sub-thread starting at
http://lists.gnu.org/archive/html/qemu-devel/2009-11/msg01071.html

#28 lore raw

Re: [Qemu-devel] Re: [FOR 0.12 PATCH 18/18] QMP: add human-readable description to error response

In reply to Luiz Capitulino
Quoted 21 lines · Luiz Capitulino
Quoted 4 lines · On Mon, 7 Dec 2009 21:37:16 +0100 Markus Armbruster <[email protected]> wrote:
QMP/qmp-spec.txt +2 -1
-{ "error": { "class": json-string, "data": json-value }, "id": json-value }
+{ "error": { "class": json-string, "data": json-value, "desc": json-string },
+  "id": json-value }

QMP/qmp-spec.txt completed because of an error condition. · line 116

Quoted 10 lines
QMP/qmp-spec.txt +2 -0
 
  Where,
 
 - The "class" member contains the error class name (eg. "ServiceUnavailable")
 - The "data" member contains specific error data and is defined in a
   per-command basis, it will be an empty json-object if the error has no data
+- The "desc" member is a human-readable error message.  Clients should
+  not attempt to parse this message.
 - The "id" member contains the transaction identification associated with
   the command execution (if issued by the Client)

As we've talked on irc, I don't agree with this change.

Quoted 3 lines

Basically, adding 'desc' to the standard error message introduces all
the problems we've discussed about free-form English strings.

It's not free form English. The 'desc' string is always autogenerated based on the error object.

It's completely redundant information because you can already generate that string, but it simplifies client creation because a lazy client does not have to include the conversion table if they only care about English error output.

Regards,

Anthony Liguori

#29 lore raw

Re: [Qemu-devel] Re: [FOR 0.12 PATCH 18/18] QMP: add human-readable description to error response

Quoted 26 lines · On Tue, 08 Dec 2009 07:18:11 -0600 Anthony Liguori <[email protected]>…
Quoted 21 lines · Luiz Capitulino
Quoted 4 lines · On Mon, 7 Dec 2009 21:37:16 +0100 Markus Armbruster <[email protected]> wrote:
QMP/qmp-spec.txt +2 -1
-{ "error": { "class": json-string, "data": json-value }, "id": json-value }
+{ "error": { "class": json-string, "data": json-value, "desc": json-string },
+  "id": json-value }

QMP/qmp-spec.txt completed because of an error condition. · line 116

Quoted 10 lines
QMP/qmp-spec.txt +2 -0
 
  Where,
 
 - The "class" member contains the error class name (eg. "ServiceUnavailable")
 - The "data" member contains specific error data and is defined in a
   per-command basis, it will be an empty json-object if the error has no data
+- The "desc" member is a human-readable error message.  Clients should
+  not attempt to parse this message.
 - The "id" member contains the transaction identification associated with
   the command execution (if issued by the Client)

As we've talked on irc, I don't agree with this change.

Quoted 3 lines

Basically, adding 'desc' to the standard error message introduces all
the problems we've discussed about free-form English strings.

Quoted 7 lines

It's not free form English. The 'desc' string is always autogenerated based on the error object.

It's completely redundant information because you can already generate that string, but it simplifies client creation because a lazy client does not have to include the conversion table if they only care about English error output.

I wonder if we could have a simpler design for the internal API if
we knew in advance that 'desc' would be part of the standard error message, eg. error code based.

But, as Markus said it's not the 'worst' as I've put it, let's
move forward then.

#30 lore raw
In reply to Markus Armbruster
Quoted 2 lines · On Mon, 7 Dec 2009 21:36:58 +0100 Markus Armbruster <[email protected]> wrote:

This patch series converts monitor commands eject, change, closefd, getfd to QError.

Thanks a lot for this work.

Although the series is already merged, I have some comments and will
reply the patches.