clipboard doesn't work

[totaam]

Issue migrated from trac ticket # 176

component: core | priority: critical | resolution: fixed

2012-08-11 09:02:21: onlyjob created the issue

---

I'm struggling to get clipboard working in 0.4.1:

On Debian Wheezy (both on server and client) copying text to clipboard (either from menu or keyboard) blanks clipboard contents of the remote machine.

Let's say I select and copy something in remote application, then trying to paste to local application: it pastes nothing and previous clipboard content disappears.

Exactly the same behaviour can be observed the other way: when I copy text to clipboard from local application, clipboard contents of remote application disappear.

Any ideas?

Thank you.

[totaam]

2012-08-11 09:26:18: antoine commented

---

Changing:

#debug = log.info
debug = log.debug

to:

debug = log.info
#debug = log.debug

in clipboard_base.py can give us a lot more debug diagnostics messages.

I will try what you describe as soon as I can (about a week).
Bear in mind that not all clipboard content types can be serialized and sent over to the other end, that some applications do weird things (Qt apps in particular IIRC), that clipboard managers simply break havoc by constantly requesting the contents of the clipboard, etc...
Which app did you have problems with? Or was this all apps?

[totaam]

2012-08-11 09:27:29: onlyjob commented

---

Apparently this is a regression because 0.4.0 works perfectly.

[totaam]

2012-08-11 09:43:07: antoine commented

---

Hmm, that's odd, all I can see that is clipboard related in 0.4.1:

• 1230 which is meant to fix x11 long/integer packing on 64-bit platforms - are both client and server 64-bit?
• 1234 which should only affect platforms without the x11 bindings (ie: win32)
• 1220 which fixes a typo which was preventing the x11 clipboard code from firing for x11 atoms

There are some further fixes due for 0.4.2:

• 1310 which should only affect platforms without the x11 bindings (ie: win32)
• r1274 and r1278 which improve keyboard support for win32

[totaam]

2012-08-11 10:45:17: onlyjob commented

---

Thanks for hints.

I reckon something went wrong in 1230 - I reverted all changes introduced to
xpra/xposix/xclipboard.py
with the attached patch and it fixed the problem.

Yes, both machines are amd64, sorry for not mentioning it.

Cheers,
Dmitry.

[totaam]

2012-08-11 10:45:56: onlyjob uploaded file fix-clipboard.patch (2.0 KiB)

patch: partial revert of r1230

[totaam]

2012-08-11 17:16:28: aelmahmoudy uploaded file fix-clipboard.diff (0.7 KiB)

[totaam]

2012-08-11 17:17:37: aelmahmoudy commented

---

I think the patch can be reduced as per fix-clipboard.diff (note: hand crafted diff file, but the idea should be clear)

[totaam]

2012-08-12 01:08:24: antoine commented

---

Hmm, there is something odd going on here.
We can't just change the struct.pack without changing the corresponding struct.unpack - this is also used in gdk_clipboard.py for atoms and fixes known bugs.

Also, according to #11 this changeset fixes a warning which does look like a size issue (see 11#comment:3)

And obviously there's the actual X11 docs, see for example:

• XGetWindowProperty: "Returns the data in the specified format. If the returned format is 8, the returned data is represented as a char array. If the returned format is 16, the returned data is represented as a array of short int type and should be cast to that type to obtain the elements. If the returned format is 32, the property data will be stored as an array of longs (which in a 64-bit application will be 64-bit values that are padded in the upper 4 bytes)."
• gdk3-Properties-and-Atoms: "A property has an associated format, an integer describing how many bits are in each unit of data inside the property. It must be 8, 16, or 32. When data is transferred between the server and client, if they are of different endianesses it will be byteswapped as necessary according to the format of the property. Note that on the client side, properties of format 32 will be stored with one unit per long, even if a long integer has more than 32 bits on the platform."
  Now the latter is confusing me slightly since it talks about padding 32 bits rather than packing into 64... I think this needs looking into by placing know values in there and seeing how they're actually represented.

[totaam]

2012-08-12 02:27:41: onlyjob commented

---

I'll be happy to help testing this issue when you realise the best way to fix it. :)

[totaam]

2012-08-12 06:53:53: antoine commented

---

Thanks - if you can provide log samples from both client and server with debug enabled for clipboard as per comment:1, that may allow me to fix the bug sooner (since I'm not seeing the problem here on my 64-bit laptop - tried both trunk and 0.4.x branch).
Also, please apply the debug change in r1314, or even better, test using trunk to also check if the bug is still present there (just in case - also the debug output in trunk shows the event time which is helpful for piecing things back together).

[totaam]

2012-08-12 09:17:07: onlyjob commented

---

Here it is:

0.4.1 (unpatched, both server and client) do NOT work:

Paste to remote application @client:

process clipboard packet type=clipboard-request
process clipboard request, request_id=0, selection=CLIPBOARD, local name=CLIPBOARD, target=TARGETS
get_contents(TARGETS,<function got_contents at 0x24a0b18>)
got_contents(ATOM,32,72)

Paste to remote application @server:

do_selection_request_event(<gtk.gdk.Event at 0x1888be8: GDK_SELECTION_REQUEST selection=CLIPBOARD, target=TARGETS, property=_QT_SELECTION>)
do_selection_get(<GtkSelectionData at 0x7fffddc50440>,0,390684085)
get clipboard from remote handler id=0
process clipboard packet type=clipboard-contents
process clipboard contents, selection=CLIPBOARD, type=ATOM, format=32
clipboard wire -> raw: ('ATOM', 32, 'atoms', ['text/plain', 'UTF8_STRING', 'STRING', 'TEXT', 'COMPOUND_TEXT', 'TARGETS', 'MULTIPLE', 'TIMESTAMP', 'SAVE_TARGETS']) -> '\xa1\x00\x00\x00G\x00\x00\x00\x1f\x00\x00\x00\xa2\x00\x00\x00\xa3\x00\x00\x00\x87\x00\x00\x00\x86\x00\x00\x00\x84\x00\x00\x00\x88\x00\x00\x00'
got clipboard contents(0)=36 (type=ATOM, format=32)
get clipboard from remote result(0)={'data': '\xa1\x00\x00\x00G\x00\x00\x00\x1f\x00\x00\x00\xa2\x00\x00\x00\xa3\x00\x00\x00\x87\x00\x00\x00\x86\x00\x00\x00\x84\x00\x00\x00\x88\x00\x00\x00', 'type': 'ATOM', 'format': 32}
do_selection_get(<GtkSelectionData at 0x7fffddc50440>,0,390684085) calling selection_data.set(ATOM,32,36)
do_selection_request_event(<gtk.gdk.Event at 0x1888be8: GDK_SELECTION_REQUEST selection=CLIPBOARD, target=COMPOUND_TEXT, property=_QT_SELECTION>)

As far as I can see all applications are affected.

Also I noticed this may be a server-side issue: if I connect to xpra 0.4.1 that I patched by reverting xpra/xposix/xclipboard.py to 0.4.0 state from unpatched 0.4.1 clipboard works both ways.

Here is a client log for such case:

Copy to clipboard @client:

do_selection_clear_event(<gtk.gdk.Event at 0x17bbfd0: GDK_SELECTION_CLEAR selection=PRIMARY, target=, property=>) selection=PRIMARY
send clipboard token: PRIMARY
do_selection_clear_event(<gtk.gdk.Event at 0x17bbfd0: GDK_SELECTION_CLEAR selection=CLIPBOARD, target=, property=>) selection=CLIPBOARD
send clipboard token: CLIPBOARD

Paste to remote application @client (works):

process clipboard packet type=clipboard-request
process clipboard request, request_id=11, selection=CLIPBOARD, local name=CLIPBOARD, target=UTF8_STRING
get_contents(UTF8_STRING,<function got_contents at 0x18517d0>)
got_contents(UTF8_STRING,8,5)

I will test trunk and report ASAP.

Cheers,
Dmitry.

[totaam]

2012-08-12 09:17:07: onlyjob

[totaam]

2012-08-12 09:43:34: onlyjob commented

---

trunk r1314 both server and client (do NOT work)

Server log (paste to remote app):

2012-08-12 18:36:57,453 do_selection_request_event(<gtk.gdk.Event at 0x3036fd0: GDK_SELECTION_REQUEST selection=CLIPBOARD, target=TARGETS, property=_QT_SELECTION>) 
2012-08-12 18:36:57,455 target for CLIPBOARD: 'TARGETS' 
2012-08-12 18:36:57,455 do_selection_request_event(<gtk.gdk.Event at 0x3036fd0: GDK_SELECTION_REQUEST selection=CLIPBOARD, target=TARGETS, property=_QT_SELECTION>) target=TARGETS, selection=CLIPBOARD 
2012-08-12 18:36:57,455 do_selection_get(<GtkSelectionData at 0x7fff10466f80>, 0, 541091) 
2012-08-12 18:36:57,455 get clipboard from remote handler id=0 
2012-08-12 18:36:57,467 process clipboard packet type=clipboard-contents 
2012-08-12 18:36:57,467 process clipboard contents, selection=CLIPBOARD, type=ATOM, format=32 
2012-08-12 18:36:57,467 _munge_wire_selection_to_raw(atoms, ATOM, 32, <type 'tuple'>:9) 
2012-08-12 18:36:57,468 _munge_wire_selection_to_raw(atoms, ATOM, 32, <type 'tuple'>:9)=[161L, 71L, 31L, 162L, 163L, 135L, 134L, 132L, 136L]=['\xa1', '\x00', '\x00', '\x00', 'G', '\x00', '\x00', '\x00', '\x1f', '\x00', '\x00', '\x00', '\xa2', '\x00', '\x00', '\x00', '\xa3', '\x00', '\x00', '\x00', '\x87', '\x00', '\x00', '\x00', '\x86', '\x00', '\x00', '\x00', '\x84', '\x00', '\x00', '\x00', '\x88', '\x00', '\x00', '\x00'] 
2012-08-12 18:36:57,468 clipboard wire -> raw: ('ATOM', 32, 'atoms', ('text/plain', 'UTF8_STRING', 'STRING', 'TEXT', 'COMPOUND_TEXT', 'TARGETS', 'MULTIPLE', 'TIMESTAMP', 'SAVE_TARGETS')) -> '\xa1\x00\x00\x00G\x00\x00\x00\x1f\x00\x00\x00\xa2\x00\x00\x00\xa3\x00\x00\x00\x87\x00\x00\x00\x86\x00\x00\x00\x84\x00\x00\x00\x88\x00\x00\x00' 
2012-08-12 18:36:57,468 got clipboard contents(0)=36 (type=ATOM, format=32) 
2012-08-12 18:36:57,468 get clipboard from remote result(0)={'data': '\xa1\x00\x00\x00G\x00\x00\x00\x1f\x00\x00\x00\xa2\x00\x00\x00\xa3\x00\x00\x00\x87\x00\x00\x00\x86\x00\x00\x00\x84\x00\x00\x00\x88\x00\x00\x00', 'type': 'ATOM', 'format': 32} 
2012-08-12 18:36:57,469 do_selection_get(<GtkSelectionData at 0x7fff10466f80>,0,541091) calling selection_data.set(ATOM, 32, <type 'str'>:36) 
2012-08-12 18:36:57,494 do_selection_request_event(<gtk.gdk.Event at 0x3036fd0: GDK_SELECTION_REQUEST selection=CLIPBOARD, target=COMPOUND_TEXT, property=_QT_SELECTION>) 
2012-08-12 18:36:57,494 target for CLIPBOARD: 'COMPOUND_TEXT' 
2012-08-12 18:36:57,495 do_selection_request_event(<gtk.gdk.Event at 0x3036fd0: GDK_SELECTION_REQUEST selection=CLIPBOARD, target=COMPOUND_TEXT, property=_QT_SELECTION>) target=COMPOUND_TEXT, selection=CLIPBOARD 

Client log (paste to remote app):

2012-08-12 18:36:57,545 process clipboard packet type=clipboard-request
2012-08-12 18:36:57,545 process clipboard request, request_id=0, selection=CLIPBOARD, local name=CLIPBOARD, target=TARGETS
2012-08-12 18:36:57,545 get_contents(TARGETS,<function got_contents at 0x358f500>)
2012-08-12 18:36:57,550 unpack(<gtk.Clipboard object at 0x2ed9730 (GtkClipboard at 0x301b340)>, <GtkSelectionData at 0x2cba1c0>, <type 'NoneType'>:0)
2012-08-12 18:36:57,550 unpack(..) type=ATOM, format=32, data=<type 'str'>:72
2012-08-12 18:36:57,551 got_contents(ATOM, 32, <type 'str'>:72) data=[143, 0, 0, 0, 0, 0, 0, 0, 71, 0, 0, 0, 0, 0, 0, 0, 31, 0, 0, 0, 0, 0, 0, 0, 144, 0, 0, 0, 0, 0, 0, 0, 121, 0, 0, 0, 0, 0, 0, 0, 138, 0, 0, 0, 0, 0, 0, 0, 140, 0, 0, 0, 0, 0, 0, 0, 141, 0, 0, 0, 0, 0, 0, 0, 142, 0, 0, 0, 0, 0, 0, 0], str(data)=�G�y����
2012-08-12 18:36:57,551 _do_munge_raw_selection_to_wire(TARGETS, ATOM, 32, <type 'str'>:72:['\x8f', '\x00', '\x00', '\x00', '\x00', '\x00', '\x00', '\x00', 'G', '\x00', '\x00', '\x00', '\x00', '\x00', '\x00', '\x00', '\x1f', '\x00', '\x00', '\x00', '\x00', '\x00', '\x00', '\x00', '\x90', '\x00', '\x00', '\x00', '\x00', '\x00', '\x00', '\x00', 'y', '\x00', '\x00', '\x00', '\x00', '\x00', '\x00', '\x00', '\x8a', '\x00', '\x00', '\x00', '\x00', '\x00', '\x00', '\x00', '\x8c', '\x00', '\x00', '\x00', '\x00', '\x00', '\x00', '\x00', '\x8d', '\x00', '\x00', '\x00', '\x00', '\x00', '\x00', '\x00', '\x8e', '\x00', '\x00', '\x00', '\x00', '\x00', '\x00', '\x00']) using gdk atom code
2012-08-12 18:36:57,552 _do_munge_raw_selection_to_wire(TARGETS, ATOM, 32, <type 'str'>:72) atoms=[<GdkAtom 0x8f # 'text/plain'>, <GdkAtom 0x47'UTF8_STRING'>, <GdkAtom 0x1f # 'STRING'>, <GdkAtom 0x90'TEXT'>, <GdkAtom 0x79 # 'COMPOUND_TEXT'>, <GdkAtom 0x8a'TARGETS'>, <GdkAtom 0x8c # 'MULTIPLE'>, <GdkAtom 0x8d'TIMESTAMP'>, <GdkAtom 0x8e = 'SAVE_TARGETS'>]
2012-08-12 18:36:57,552 _do_munge_raw_selection_to_wire(TARGETS, ATOM, 32, <type 'str'>:72) atom_names=['text/plain', 'UTF8_STRING', 'STRING', 'TEXT', 'COMPOUND_TEXT', 'TARGETS', 'MULTIPLE', 'TIMESTAMP', 'SAVE_TARGETS']
2012-08-12 18:36:57,552 clipboard raw -> wire: ('ATOM', 32, '\x8f\x00\x00\x00\x00\x00\x00\x00G\x00\x00\x00\x00\x00\x00\x00\x1f\x00\x00\x00\x00\x00\x00\x00\x90\x00\x00\x00\x00\x00\x00\x00y\x00\x00\x00\x00\x00\x00\x00\x8a\x00\x00\x00\x00\x00\x00\x00\x8c\x00\x00\x00\x00\x00\x00\x00\x8d\x00\x00\x00\x00\x00\x00\x00\x8e\x00\x00\x00\x00\x00\x00\x00') -> ('atoms', ['text/plain', 'UTF8_STRING', 'STRING', 'TEXT', 'COMPOUND_TEXT', 'TARGETS', 'MULTIPLE', 'TIMESTAMP', 'SAVE_TARGETS'])

[totaam]

2012-08-12 10:12:47: antoine commented

---

thanks.
this sample is from a qt app, right? do you get the same problem with an xterm?
had problems with qt before, see #84

it also looks like the atoms aren't being filtered, maybe because the target does not match? here - does it fix the problem if you force the atom filtering to apply?
you want to remove "COMPOUND_TEXT" and "SAVE_TARGETS".

they may have got removed accidentally by using the "wrong" code previously if those were ending up in the highest 32 bits.

[totaam]

2012-08-12 10:12:47: antoine