1:54 p.m.
http://bugzilla.moblin.org/show_bug.cgi?id=5043
Summary: move command line into D-Bus server
Classification: Moblin Projects
Product: SyncEvolution
Version: upstream
Platform: Generic
OS/Version: Other
Status: ASSIGNED
Severity: normal
Priority: P2
Component: * Feature Request
AssignedTo: syncevolution(a)lists.intel.com
ReportedBy: patrick.ohly(a)intel.com
CC: syncevolution(a)lists.intel.com
Depends on: 5041,5042
With output redirection working (#5041) and logging sent to clients (#5042) it
should be fairly easy to just pass all of argv to a new D-Bus method and have
it processed as part of the syncevo-dbus-server. In that mode, syncevolution is
really just another client of the single instance which controls syncs.
Care has to be taken to properly serialize requests by syncevolution with
on-going operations.
A new command line option "--monitor" would be useful to connect to a running
sync session. "--status" without server config parameter could be used to query
whether there is any sync running.
--
Configure bugmail: http://bugzilla.moblin.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are watching the assignee of the bug.
You are watching someone on the CC list of the bug.
12:22 p.m.
New subject: [Bug 5043] move command line into D-Bus server
http://bugzilla.moblin.org/show_bug.cgi?id=5043
pohly <patrick.ohly(a)intel.com> changed:
What |Removed |Added
----------------------------------------------------------------------------
Blocks| |7210
--
Configure bugmail: http://bugzilla.moblin.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are watching the assignee of the bug.
You are watching someone on the CC list of the bug.
8:58 a.m.
New subject: [Bug 5043] move command line into D-Bus server
http://bugzilla.moblin.org/show_bug.cgi?id=5043
yongsheng zhu <yongsheng.zhu(a)intel.com> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |yongsheng.zhu(a)intel.com
--- Comment #1 from yongsheng zhu <yongsheng.zhu(a)intel.com> 2010-03-10 00:58:04 PST
---
1)Class Cmdline is used to do sync, configure, etc. The first idea is to modify
it in Cmdline.h/cpp, however, it belongs to libsyncevolution and now cppunit
tests are based on it, so i think it isn't suitable to use it in
syncevolution.cpp. New command line should not use libsyncevolution. so I
suggest moving it out of libsyncevolution. Currently we can leave it there not
to interfere with nightly tests.
2)For the option '--version' in the command line, dbus server don't support
it.
I suggest a new method in dbus api 'GetCapability' returning a list of strings
to represent server related info.
3) when no any parameters in command line, current behavior is to print
registered sources' databases. Now dbus api don't have this kind of API. Is it
ok to add a new API for this?
4) '--keyring' is tried to be removed since dbus server makes the decision.
5) DBus server has no api to return valid properties and valid property values.
do we really need this? I think yes, it would be better to add new APIs.
6) DBus server has no api to know all backends info, e.g, info from
SyncSource::backendsInfo(), also through 'GetCapability'?
7)'restore' method don't support 'dryrun' parameter in current DBus
API. Add
it?
--
Configure bugmail: http://bugzilla.moblin.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are watching the assignee of the bug.
You are watching someone on the CC list of the bug.
2:04 p.m.
New subject: [Bug 5043] move command line into D-Bus server
http://bugzilla.moblin.org/show_bug.cgi?id=5043
--- Comment #2 from pohly <patrick.ohly(a)intel.com> 2010-03-10 06:04:05 PST ---
(In reply to comment #1)
1)Class Cmdline is used to do sync, configure, etc. The first idea is
to modify
it in Cmdline.h/cpp, however, it belongs to libsyncevolution and now cppunit
tests are based on it, so i think it isn't suitable to use it in
syncevolution.cpp. New command line should not use libsyncevolution.
I disagree on that. The new command line tool should be able to do both: run as
small wrapper around D-Bus *and* call Cmdline/libsyncevolution.so directly. The
latter is useful a) when SyncEvolution is compiled without D-Bus and b) for
SyncEvolution client<->SyncEvolution server tests.
Right now, the later won't work when the command line uses the D-Bus server
because the D-Bus server only supports one session at a time.
2)For the option '--version' in the command line, dbus server
don't support it.
I suggest a new method in dbus api 'GetCapability' returning a list of strings
to represent server related info.
I think this is going in the wrong direction. You seem to attempt to
reimplement all command line options as D-Bus operations. This is not what I
meant with "pass all argv to a new D-Bus method and have it processed as part
of syncevo-dbus-server".
In that all approach, all that is needed is one new API: "execute(char
*argv[])". It'll get a D-Bus session, then inside that session invoke the
Cmdline methods, which can call the same code as they do now instead of going
through the D-Bus API.
The main gaps around that are output handling and integrating the session
started by "--run" with the current D-Bus Session (not sure whether that is
necessary, but I suspect so).
3) when no any parameters in command line, current behavior is to
print
registered sources' databases. Now dbus api don't have this kind of API. Is it
ok to add a new API for this?
4) '--keyring' is tried to be removed since dbus server makes the decision.
It would still be needed for "non-D-Bus mode".
5) DBus server has no api to return valid properties and valid
property values.
do we really need this? I think yes, it would be better to add new APIs.
See 2).
6) DBus server has no api to know all backends info, e.g, info from
SyncSource::backendsInfo(), also through 'GetCapability'?
See 2)
7)'restore' method don't support 'dryrun'
parameter in current DBus API. Add
it?
We might want to add it purely to have it feature-complete.
--
Configure bugmail: http://bugzilla.moblin.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are watching the assignee of the bug.
You are watching someone on the CC list of the bug.
1:57 a.m.
New subject: [Bug 5043] move command line into D-Bus server
http://bugzilla.moblin.org/show_bug.cgi?id=5043
--- Comment #3 from yongsheng zhu <yongsheng.zhu(a)intel.com> 2010-03-10 17:57:18 PST
---
(In reply to comment #2)
(In reply to comment #1)
> 1)Class Cmdline is used to do sync, configure, etc. The first idea is to modify
> it in Cmdline.h/cpp, however, it belongs to libsyncevolution and now cppunit
> tests are based on it, so i think it isn't suitable to use it in
> syncevolution.cpp. New command line should not use libsyncevolution.
I disagree on that. The new command line tool should be able to do both: run as
small wrapper around D-Bus *and* call Cmdline/libsyncevolution.so directly. The
latter is useful a) when SyncEvolution is compiled without D-Bus and b) for
SyncEvolution client<->SyncEvolution server tests.
Right now, the later won't work when the command line uses the D-Bus server
because the D-Bus server only supports one session at a time.
> 2)For the option '--version' in the command line, dbus server don't
support it.
> I suggest a new method in dbus api 'GetCapability' returning a list of
strings
> to represent server related info.
I think this is going in the wrong direction. You seem to attempt to
reimplement all command line options as D-Bus operations. This is not what I
meant with "pass all argv to a new D-Bus method and have it processed as part
of syncevo-dbus-server".
In that all approach, all that is needed is one new API: "execute(char
*argv[])". It'll get a D-Bus session, then inside that session invoke the
Cmdline methods, which can call the same code as they do now instead of going
through the D-Bus API.
The main gaps around that are output handling and integrating the session
started by "--run" with the current D-Bus Session (not sure whether that is
necessary, but I suspect so).
Yes, get your idea. I'll add a method in
org.syncevolution.Server. Naturally,
to listen to outputs of Cmdline, we have to solve #5041 and #5042.
--
Configure bugmail: http://bugzilla.moblin.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are watching the assignee of the bug.
You are watching someone on the CC list of the bug.
8:25 a.m.
New subject: [Bug 5043] move command line into D-Bus server
http://bugzilla.moblin.org/show_bug.cgi?id=5043
yongsheng zhu <yongsheng.zhu(a)intel.com> changed:
What |Removed |Added
----------------------------------------------------------------------------
AssignedTo|syncevolution(a)lists.intel.c |yongsheng.zhu(a)intel.com
|om |
--- Comment #4 from yongsheng zhu <yongsheng.zhu(a)intel.com> 2010-03-11 00:25:37 PST
---
Not only doing sync should be integrated to DBus Session, but also many other
operations, like restore, setConfig, remove, migrate. It seems much more
complex to integrate them and may break current working logic. Is it ok to
simply the logic to handle command line request only when there is no session
in the working queue? It should not. Because dbus server can't be aware it.
but what if the created session is put in the working queue? command line users
waiting for this? I suggest below trade-off behaviors:
1) For those opeartions, like 'version', 'GetConfigs', who don't need
an active
session, let cmdline processes them directly and redirect its log output and
sent them with dbus signals to client.
2) For others, depending on active sessions,if no other sessions, create a
session and run with the args.
3) else if there are already other sessions, returns directly with a warning
since command line users should not wait. However, even this assumption, it has
to be integrated into session for there might be other requests when doing
things for command line request.
--
Configure bugmail: http://bugzilla.moblin.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are watching the assignee of the bug.
You are watching someone on the CC list of the bug.
1:57 p.m.
New subject: [Bug 5043] move command line into D-Bus server
http://bugzilla.moblin.org/show_bug.cgi?id=5043
--- Comment #5 from pohly <patrick.ohly(a)intel.com> 2010-03-11 05:57:43 PST ---
(In reply to comment #4)
Not only doing sync should be integrated to DBus Session, but also
many other
operations, like restore, setConfig, remove, migrate.
Of course. I was thinking that we would always execute inside a session, to be
on the safe side. Detecting read-only operations would be nice, but I don't see
that as essential.
It seems much more
complex to integrate them and may break current working logic. Is it ok to
simply the logic to handle command line request only when there is no session
in the working queue?
Is that really easier? How would you ensure that a command line request gets
executed before a different request that came later?
I suggest below trade-off behaviors:
1) For those opeartions, like 'version', 'GetConfigs', who don't need
an active
session, let cmdline processes them directly and redirect its log output and
sent them with dbus signals to client.
2) For others, depending on active sessions,if no other sessions, create a
session and run with the args.
3) else if there are already other sessions, returns directly with a warning
since command line users should not wait.
We can't just abort the request in case 3. The command line "user" might be
an
application like the Maemo GUI or the current Genesis GUI, in which case it is
fine to have them wait. An "INFO" message would be nice, but because this is
going to happen rarely, we shouldn't invest too much effort into it.
Even the command line user would prefer that. Because what would he do when the
server rejects the request? All he can do is submit it again, hoping for better
luck the next time.
--
Configure bugmail: http://bugzilla.moblin.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are watching someone on the CC list of the bug.
7:34 a.m.
New subject: [Bug 5043] move command line into D-Bus server
http://bugzilla.moblin.org/show_bug.cgi?id=5043
--- Comment #6 from pohly <patrick.ohly(a)intel.com> 2010-03-17 00:34:17 PST ---
CTRL-C handling needs to be implemented inside syncevolution.cpp when it uses
the D-Bus server. It should to be mapped to the corresponding Suspend/Abort
D-Bus calls, to emulate the current behavior.
That leaves the question whether it should be possible to shut down the command
line while keeping the sync running. Might be useful, but not useful enough to
invest extra effort IMHO.
--
Configure bugmail: http://bugzilla.moblin.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are watching someone on the CC list of the bug.
8:37 a.m.
New subject: [Bug 5043] move command line into D-Bus server
http://bugzilla.moblin.org/show_bug.cgi?id=5043
--- Comment #7 from yongsheng zhu <yongsheng.zhu(a)intel.com> 2010-03-18 01:37:20 PST
---
A new command line option "--monitor" would be useful to
connect to a running
sync session. "--status" without server config parameter could be used to
query
whether there is any sync running.
--status: iterate all sessions and check them by calling 'Session.getStatus'
and check status is 'running'. If a session is not found, message "No sync
is
running" is printed. If session is found, message "'configname' is
running".
--monitor: first check whether there is any sync running by the above method.
connect to a running sync session and listen to its status changes and print
them until it is completed.
reasonable?
--
Configure bugmail: http://bugzilla.moblin.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are watching someone on the CC list of the bug.
9:25 a.m.
New subject: [Bug 5043] move command line into D-Bus server
http://bugzilla.moblin.org/show_bug.cgi?id=5043
--- Comment #8 from pohly <patrick.ohly(a)intel.com> 2010-03-18 02:25:31 PST ---
(In reply to comment #7)
> A new command line option "--monitor" would be useful
to connect to a running
> sync session. "--status" without server config parameter could be used to
query
> whether there is any sync running.
--status: iterate all sessions and check them by calling 'Session.getStatus'
and check status is 'running'. If a session is not found, message "No sync
is
running" is printed. If session is found, message "'configname' is
running".
--monitor: first check whether there is any sync running by the above method.
connect to a running sync session and listen to its status changes and print
them until it is completed.
reasonable?
Conceptually, more than one session might be running. Even though it cannot
happen right now (limitation of the syncevo-dbus-server), we should prepare the
client for it, if it is not too much work.
For --status that means that it should print "running" for all sessions. For
--monitor we should accept an optional config name. If one is given, only
attach to a session of that config, otherwise print an error "'configname'
is
not running". If none is given, pick the first.
Careful with config name comparisons: "ScheduleWorld" on the command line might
expand to a config "scheduleworld@other-context", which is what will be
reported by GetStatus(). Because the syncevolution binary is linked against
libsyncevolution, it can use the SyncConfig class to find the real config.
There's no D-Bus call for that.
--
Configure bugmail: http://bugzilla.moblin.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are watching someone on the CC list of the bug.
2:50 a.m.
New subject: [Bug 5043] move command line into D-Bus server
http://bugzilla.moblin.org/show_bug.cgi?id=5043
--- Comment #9 from yongsheng zhu <yongsheng.zhu(a)intel.com> 2010-03-23 19:50:52 PST
---
For
--monitor we should accept an optional config name. If one is given,
only
attach to a session of that config, otherwise print an error "'configname'
is
not running". If none is given, pick the first.
For --monitor, what kind of
information should be printed in the console? Print
info of 'LogOutput' signals?
--
Configure bugmail: http://bugzilla.moblin.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are watching someone on the CC list of the bug.
7:29 a.m.
New subject: [Bug 5043] move command line into D-Bus server
http://bugzilla.moblin.org/show_bug.cgi?id=5043
--- Comment #10 from pohly <patrick.ohly(a)intel.com> 2010-03-24 00:29:37 PST ---
(In reply to comment #9)
For
> --monitor we should accept an optional config name. If one is given, only
> attach to a session of that config, otherwise print an error
"'configname' is
> not running". If none is given, pick the first.
For --monitor, what kind of information should be printed in the console? Print
info of 'LogOutput' signals?
Yes.
--
Configure bugmail: http://bugzilla.moblin.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are watching someone on the CC list of the bug.
6:04 p.m.
New subject: [Bug 5043] move command line into D-Bus server
http://bugzilla.moblin.org/show_bug.cgi?id=5043
--- Comment #12 from pohly <patrick.ohly(a)intel.com> 2010-03-26 11:04:42 PST ---
"syncevolution foo bar" fails (non-zero return code) without showing any output
when using the daemon.
--
Configure bugmail: http://bugzilla.moblin.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are watching someone on the CC list of the bug.
1:26 a.m.
New subject: [Bug 5043] move command line into D-Bus server
http://bugzilla.moblin.org/show_bug.cgi?id=5043
--- Comment #23 from yongsheng zhu <yongsheng.zhu(a)intel.com> 2010-04-08 18:26:25
PST ---
(In reply to comment #22)
(In reply to comment #21)
> the patch for 'password hang' is ready on the branch 'mb5043'.
please review
> and merge.
As discussed by chat, the change of the InfoRequest prototype needs to be
tested with the sync-ui, which Yongsheng volunteered to do.
it seems that my two
machines have the same problem on dbus and glib. I start
sync-ui(built from master, not my branch), but it tells me that
"(sync-ui:25464): GLib-GObject-WARNING **:
/build/buildd/glib2.0-2.22.3/gobject/gsignal.c:1563: parameter 6 of type
`<invalid>' for signal "SyncevoServer::info_request" is not a value
type"
thus sync-ui can't listen to InfoRequest signal.
Since the handler will cause dbus server exiting, I doubt they are because my
dbus and glib library but find nothing by referencing Congwu's environment. so
firstly I'll commit the patch without changing the InfoRequest's arguments.
--
Configure bugmail: http://bugzilla.moblin.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are watching someone on the CC list of the bug.
1:50 a.m.
New subject: [Bug 5043] move command line into D-Bus server
http://bugzilla.moblin.org/show_bug.cgi?id=5043
--- Comment #24 from yongsheng zhu <yongsheng.zhu(a)intel.com> 2010-04-08 18:50:54
PST ---
sync-ui and dbus-server in the master are using 'string' as the type of
'handler'. so i don't want to change them back to the 'object' type.
Anyone who
can help have a test for sync ui since my environment is blocked by the signal
problem?
I'll also update the commit message.
--
Configure bugmail: http://bugzilla.moblin.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are watching someone on the CC list of the bug.
2:13 a.m.
New subject: [Bug 5043] move command line into D-Bus server
http://bugzilla.moblin.org/show_bug.cgi?id=5043
--- Comment #25 from yongsheng zhu <yongsheng.zhu(a)intel.com> 2010-04-08 19:13:47
PST ---
the change is done. thanks.
--
Configure bugmail: http://bugzilla.moblin.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are watching someone on the CC list of the bug.
6:56 a.m.
New subject: [Bug 5043] move command line into D-Bus server
http://bugzilla.moblin.org/show_bug.cgi?id=5043
pohly <patrick.ohly(a)intel.com> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|REOPENED |RESOLVED
Resolution| |FIXED
--- Comment #26 from pohly <patrick.ohly(a)intel.com> 2010-04-08 23:56:46 PST ---
(In reply to comment #25)
the change is done. thanks.
Merged. In the commit message for the .xml type fix I added a comment
explaining why we use string and not object path - basically because we are
unsure about "" being a valid object path.
What remains is the error in comment #25. I'll open a new issue.
--
Configure bugmail: http://bugzilla.moblin.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are watching someone on the CC list of the bug.
5:18 p.m.
New subject: [Bug 5043] move command line into D-Bus server
http://bugzilla.moblin.org/show_bug.cgi?id=5043
pohly <patrick.ohly(a)intel.com> changed:
What |Removed |Added
----------------------------------------------------------------------------
Resolution|FIXED |INTEGRATED
--- Comment #27 from pohly <patrick.ohly(a)intel.com> 2010-04-15 10:18:11 PST ---
In MeeGo builds 0.9.99.19 or 0.9.99.20.
--
Configure bugmail: http://bugzilla.moblin.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are watching someone on the CC list of the bug.
1:53 a.m.
New subject: [Bug 5043] move command line into D-Bus server
http://bugzilla.moblin.org/show_bug.cgi?id=5043
ZhangJingke <jingke.zhang(a)intel.com> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|RESOLVED |VERIFIED
CC| |jingke.zhang(a)intel.com
--- Comment #28 from ZhangJingke <jingke.zhang(a)intel.com> 2010-04-15 18:53:15 PST
---
Yes, currently syncevolution commands are using 'daemon=yes' by default from
0.9.99.19.
Verify it now.
--
Configure bugmail: http://bugzilla.moblin.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are watching someone on the CC list of the bug.
4547
days inactive
4802
days old
syncevolution-issues@syncevolution.org
18 comments
1 participants
participants (1)
-
bugzilla＠moblin.org