"pasteboard doesn't contain valid data" on OSX

"pasteboard doesn't contain valid data" on OSX

[Randal L. Schwartz]

This is recently broken.  "pasteboard doesn't contain valid data"
on any use of ^Y paste after any kind of cut or copy within Emacs.

Pasting *does* seem to work if the text came from another application
though, so it's more likely the copy/cut operation in emacs.

I haven't had time to bisect yet, but hopefully this will ring a bell
for someone to go investigate.

-- 
Randal L. Schwartz - Stonehenge Consulting Services, Inc. - +1 503 777 0095
<merlyn@stonehenge.com> <URL:http://www.stonehenge.com/merlyn/>
Smalltalk/Perl/Unix consulting, Technical writing, Comedy, etc. etc.
See http://methodsandmessages.vox.com/ for Smalltalk and Seaside discussion

Re: "pasteboard doesn't contain valid data" on OSX

[Jan Djärv]

[-- Attachment #1: Type: text/plain, Size: 573 bytes --]

It is known.  Part of the selection-changes being done.
The attached patch fixes it (nsselect.m changes by David De La Harpe Golden).

	Jan D.


Randal L. Schwartz skrev 2010-08-07 21.11:
>
> This is recently broken.  "pasteboard doesn't contain valid data"
> on any use of ^Y paste after any kind of cut or copy within Emacs.
>
> Pasting *does* seem to work if the text came from another application
> though, so it's more likely the copy/cut operation in emacs.
>
> I haven't had time to bisect yet, but hopefully this will ring a bell
> for someone to go investigate.
>

[-- Attachment #2: clipboard.diff --]
[-- Type: text/plain, Size: 2591 bytes --]

=== modified file 'lisp/term/ns-win.el'
--- lisp/term/ns-win.el	2010-07-13 10:41:49 +0000
+++ lisp/term/ns-win.el	2010-08-06 10:52:23 +0000
@@ -1003,7 +1003,7 @@
 
 (defun ns-get-pasteboard ()
   "Returns the value of the pasteboard."
-  (ns-get-cut-buffer-internal 'PRIMARY))
+  (ns-get-cut-buffer-internal 'CLIPBOARD))
 
 (declare-function ns-store-cut-buffer-internal "nsselect.m" (buffer string))
 
@@ -1011,7 +1011,7 @@
   "Store STRING into the pasteboard of the Nextstep display server."
   ;; Check the data type of STRING.
   (if (not (stringp string)) (error "Nonstring given to pasteboard"))
-  (ns-store-cut-buffer-internal 'PRIMARY string))
+  (ns-store-cut-buffer-internal 'CLIPBOARD string))
 
 ;; We keep track of the last text selected here, so we can check the
 ;; current selection against it, and avoid passing back our own text

=== modified file 'src/nsselect.m'
--- src/nsselect.m	2010-07-08 21:25:08 +0000
+++ src/nsselect.m	2010-08-06 10:46:52 +0000
@@ -36,7 +36,7 @@
 
 #define CUT_BUFFER_SUPPORT
 
-Lisp_Object QPRIMARY, QSECONDARY, QTEXT, QFILE_NAME;
+Lisp_Object QCLIPBOARD, QPRIMARY, QSECONDARY, QTEXT, QFILE_NAME;
 
 static Lisp_Object Vns_sent_selection_hooks;
 static Lisp_Object Vns_lost_selection_hooks;
@@ -45,6 +45,8 @@
 
 static Lisp_Object Qforeign_selection;
 
+/* NSGeneralPboard is pretty much analogous to X11 CLIPBOARD */
+NSString *NXPrimaryPboard;
 NSString *NXSecondaryPboard;
 
 
@@ -60,7 +62,8 @@
 symbol_to_nsstring (Lisp_Object sym)
 {
   CHECK_SYMBOL (sym);
-  if (EQ (sym, QPRIMARY))     return NSGeneralPboard;
+  if (EQ (sym, QCLIPBOARD))     return NSGeneralPboard;
+  if (EQ (sym, QPRIMARY))     return NXPrimaryPboard;
   if (EQ (sym, QSECONDARY))   return NXSecondaryPboard;
   if (EQ (sym, QTEXT))        return NSStringPboardType;
   return [NSString stringWithUTF8String: SDATA (XSYMBOL (sym)->xname)];
@@ -71,6 +74,8 @@
 ns_string_to_symbol (NSString *t)
 {
   if ([t isEqualToString: NSGeneralPboard])
+    return QCLIPBOARD;
+  if ([t isEqualToString: NXPrimaryPboard])
     return QPRIMARY;
   if ([t isEqualToString: NXSecondaryPboard])
     return QSECONDARY;
@@ -536,12 +541,14 @@
 void
 nxatoms_of_nsselect (void)
 {
-  NXSecondaryPboard = @"Selection";
+  NXPrimaryPboard = @"PrimarySelection";
+  NXSecondaryPboard = @"SecondarySelection";
 }
 
 void
 syms_of_nsselect (void)
 {
+  QCLIPBOARD = intern ("CLIPBOARD");	staticpro (&QCLIPBOARD);
   QPRIMARY   = intern ("PRIMARY");	staticpro (&QPRIMARY);
   QSECONDARY = intern ("SECONDARY");	staticpro (&QSECONDARY);
   QTEXT      = intern ("TEXT"); 	staticpro (&QTEXT);

Re: "pasteboard doesn't contain valid data" on OSX

[David De La Harpe Golden]

On 08/08/10 20:17, Jan Djärv wrote:
> It is known.  Part of the selection-changes being done.
> The attached patch fixes it (nsselect.m changes by David De La Harpe
> Golden).
>

Eep. No it doesn't, as noted under #6677, and that's an earlier version 
too, doesn't take into account the subsequent discovery of pasteboard 
names gnustep/x11's pasteboard daemon apparently uses for PRIMARY 
("Selection") and SECONDARY ("Secondary"). (CLIPBOARD of course being 
General Pasteboard).

Plus, the ns port is doing something strange elsewhere (ns-win.el) (that 
can be summed up as "cut buffers? ...wtf?").

I was actually poking at it just now prompted by Randal's email, though 
I'm mostly still trying to catch up to the latest round of changes to 
trunk in the general area so I can (constructively) criticise them, my 
personal priority is x11 > ns > w32.

Re: "pasteboard doesn't contain valid data" on OSX

[David De La Harpe Golden]

[-- Attachment #1: Type: text/plain, Size: 1076 bytes --]

On 08/08/10 22:07, David De La Harpe Golden wrote:
>
> Plus, the ns port is doing something strange elsewhere (ns-win.el) (that
> can be summed up as "cut buffers? ...wtf?").
>

Seems stuff that's named "cut buffer" in the ns-specific parts has 
basically nothing to do with x11 cut buffers.  I don't know if there was 
a misunderstanding, or an active decision to reuse the name, or what (I 
suppose I could paw through the logs), but it's a terrible name given 
the opportunity for confusion with an obsolete x11 feature (especially 
remembering that ns emacs /runs on/ x11 in some incarnations).

Anyway. The attached patch brings emacs to the point I can call the 
(unfortunately named) ns-store/get-cut-buffer-internal functions* in ns 
emacs to set/get the various ns pasteboards that gnustep/x11 maps to 
clipboard/primary/secondary as previously mentioned, and read/write them 
from x11 selections in x11 emacs etc.

It should also improve the macosx experience (untested), though there 
are still problems*.

* N.B., higher level stuff isn't all working right yet.



[-- Attachment #2: ns-clipboard-pboard_r2.diff --]
[-- Type: text/x-patch, Size: 3769 bytes --]

=== modified file 'lisp/term/ns-win.el'
--- lisp/term/ns-win.el	2010-07-13 10:41:49 +0000
+++ lisp/term/ns-win.el	2010-08-08 22:40:49 +0000
@@ -998,25 +998,25 @@
 
 
 ;;;; Pasteboard support.
 
 (declare-function ns-get-cut-buffer-internal "nsselect.m" (buffer))
 
 (defun ns-get-pasteboard ()
   "Returns the value of the pasteboard."
-  (ns-get-cut-buffer-internal 'PRIMARY))
+  (ns-get-cut-buffer-internal 'CLIPBOARD))
 
 (declare-function ns-store-cut-buffer-internal "nsselect.m" (buffer string))
 
 (defun ns-set-pasteboard (string)
   "Store STRING into the pasteboard of the Nextstep display server."
   ;; Check the data type of STRING.
   (if (not (stringp string)) (error "Nonstring given to pasteboard"))
-  (ns-store-cut-buffer-internal 'PRIMARY string))
+  (ns-store-cut-buffer-internal 'CLIPBOARD string))
 
 ;; We keep track of the last text selected here, so we can check the
 ;; current selection against it, and avoid passing back our own text
 ;; from x-cut-buffer-or-selection-value.
 (defvar ns-last-selected-text nil)
 
 (defun x-select-text (text &optional push)
   "Select TEXT, a string, according to the window system.

=== modified file 'src/nsselect.m'
--- src/nsselect.m	2010-07-08 21:25:08 +0000
+++ src/nsselect.m	2010-08-08 20:42:05 +0000
@@ -31,51 +31,56 @@
 #include <setjmp.h>
 
 #include "lisp.h"
 #include "nsterm.h"
 #include "termhooks.h"
 
 #define CUT_BUFFER_SUPPORT
 
-Lisp_Object QPRIMARY, QSECONDARY, QTEXT, QFILE_NAME;
+Lisp_Object QCLIPBOARD, QPRIMARY, QSECONDARY, QTEXT, QFILE_NAME;
 
 static Lisp_Object Vns_sent_selection_hooks;
 static Lisp_Object Vns_lost_selection_hooks;
 static Lisp_Object Vselection_alist;
 static Lisp_Object Vselection_converter_alist;
 
 static Lisp_Object Qforeign_selection;
 
-NSString *NXSecondaryPboard;
 
+/* CLIPBOARD mapped to NSGeneralPboard */
+NSString *NXPrimaryPboard;
+NSString *NXSecondaryPboard;
 
 
 /* ==========================================================================
 
     Internal utility functions
 
    ========================================================================== */
 
 
 static NSString *
 symbol_to_nsstring (Lisp_Object sym)
 {
   CHECK_SYMBOL (sym);
-  if (EQ (sym, QPRIMARY))     return NSGeneralPboard;
+  if (EQ (sym, QCLIPBOARD))   return NSGeneralPboard;
+  if (EQ (sym, QPRIMARY))     return NXPrimaryPboard;
   if (EQ (sym, QSECONDARY))   return NXSecondaryPboard;
   if (EQ (sym, QTEXT))        return NSStringPboardType;
   return [NSString stringWithUTF8String: SDATA (XSYMBOL (sym)->xname)];
 }
 
 
 static Lisp_Object
 ns_string_to_symbol (NSString *t)
 {
   if ([t isEqualToString: NSGeneralPboard])
+    return QCLIPBOARD;
+  if ([t isEqualToString: NXPrimaryPboard])
     return QPRIMARY;
   if ([t isEqualToString: NXSecondaryPboard])
     return QSECONDARY;
   if ([t isEqualToString: NSStringPboardType])
     return QTEXT;
   if ([t isEqualToString: NSFilenamesPboardType])
     return QFILE_NAME;
   if ([t isEqualToString: NSTabularTextPboardType])
@@ -531,22 +536,27 @@
   return Qnil;
 }
 #endif
 
 
 void
 nxatoms_of_nsselect (void)
 {
-  NXSecondaryPboard = @"Selection";
+  /* These are the names GNUStep's pasteboard daemon uses,
+     using them for interoperability */
+
+  NXPrimaryPboard = @"Selection";
+  NXSecondaryPboard = @"Secondary";
 }
 
 void
 syms_of_nsselect (void)
 {
+  QCLIPBOARD = intern ("CLIPBOARD");	staticpro (&QCLIPBOARD);
   QPRIMARY   = intern ("PRIMARY");	staticpro (&QPRIMARY);
   QSECONDARY = intern ("SECONDARY");	staticpro (&QSECONDARY);
   QTEXT      = intern ("TEXT"); 	staticpro (&QTEXT);
   QFILE_NAME = intern ("FILE_NAME"); 	staticpro (&QFILE_NAME);
 
   defsubr (&Sx_disown_selection_internal);
   defsubr (&Sx_get_selection_internal);
   defsubr (&Sx_own_selection_internal);

Re: "pasteboard doesn't contain valid data" on OSX

[Jan Djärv]

David De La Harpe Golden skrev 2010-08-08 23.07:
> On 08/08/10 20:17, Jan Djärv wrote:
>> It is known. Part of the selection-changes being done.
>> The attached patch fixes it (nsselect.m changes by David De La Harpe
>> Golden).
>>
>
> Eep. No it doesn't, as noted under #6677, and that's an earlier version too,
> doesn't take into account the subsequent discovery of pasteboard names
> gnustep/x11's pasteboard daemon apparently uses for PRIMARY ("Selection") and
> SECONDARY ("Secondary"). (CLIPBOARD of course being General Pasteboard).

I run Emacs on OSX every day and I do test things before posting them.  It 
fixes it in the sense that M-w C-y works and copy/paste to/from other 
applications also work.  As 6677 sais, I did modify ns-term.el.  I assume 
there are other things that doesn't work, but at least this gives a workable 
Emacs while waiting for the real fix.

	Jan D.

Re: "pasteboard doesn't contain valid data" on OSX

[Adrian Robert]

David De La Harpe Golden <david <at> harpegolden.net> writes:

> > Plus, the ns port is doing something strange elsewhere (ns-win.el) (that
> > can be summed up as "cut buffers? ...wtf?").
> >
> 
> Seems stuff that's named "cut buffer" in the ns-specific parts has 
> basically nothing to do with x11 cut buffers.  I don't know if there was 
> a misunderstanding, or an active decision to reuse the name, or what (I 
> suppose I could paw through the logs), but it's a terrible name given 
> the opportunity for confusion with an obsolete x11 feature (especially 
> remembering that ns emacs /runs on/ x11 in some incarnations).


Please feel free to rename as seems appropriate.  Likely the naming is left over
from an older incarnation of emacs internals that got modified in other ports
but not here, or from some internal evolution of the NS code itself that kept
the old names but changed the implementations.

-Adrian

Re: "pasteboard doesn't contain valid data" on OSX

[David De La Harpe Golden]

On 09/08/10 07:25, Jan Djärv wrote:


> I assume
> there are other things that doesn't work, but at least this gives a
> workable Emacs while waiting for the real fix.
>

Yes there are, but since the latest changes to the mechanism of 
select-active-regions on trunk, they're more subtle than they were. On 
the whole you're right, my bad, sorry.

Re: "pasteboard doesn't contain valid data" on OSX

[Randal L. Schwartz]

>>>>> "Jan" == Jan Djärv <jan.h.d@swipnet.se> writes:

Jan> I run Emacs on OSX every day and I do test things before posting them.  It
Jan> fixes it in the sense that M-w C-y works and copy/paste to/from other
Jan> applications also work.  As 6677 sais, I did modify ns-term.el.  I assume
Jan> there are other things that doesn't work, but at least this gives a workable
Jan> Emacs while waiting for the real fix.

Still broken for me.  Repeat by switch-to-buffer TMP, type a line of
text, select all, ESC w, ^Y.  "pasteboard" error.

Oddly enough, it *does* work ok for ^K kills.  Not sure why those go
through two different paths.

-- 
Randal L. Schwartz - Stonehenge Consulting Services, Inc. - +1 503 777 0095
<merlyn@stonehenge.com> <URL:http://www.stonehenge.com/merlyn/>
Smalltalk/Perl/Unix consulting, Technical writing, Comedy, etc. etc.
See http://methodsandmessages.vox.com/ for Smalltalk and Seaside discussion

Re: "pasteboard doesn't contain valid data" on OSX

[Jan Djärv]

Randal L. Schwartz skrev 2010-08-11 16.23:

>
> Still broken for me.  Repeat by switch-to-buffer TMP, type a line of
> text, select all, ESC w, ^Y.  "pasteboard" error.
>
> Oddly enough, it *does* work ok for ^K kills.  Not sure why those go
> through two different paths.

Works for me with the patch I sent.  But that patch isn't checked in.  You did 
recompile and install ns-win.el didn't you?

	Jan D.

Re: "pasteboard doesn't contain valid data" on OSX

[Randal L. Schwartz]

>>>>> "Jan" == Jan Djärv <jan.h.d@swipnet.se> writes:

Jan> Works for me with the patch I sent.  But that patch isn't checked in.  You did
Jan> recompile and install ns-win.el didn't you?

I'm doing a make bootstrap after checking out HEAD... which seems to
recompile a lot of files, so I'm guessing yes.

-- 
Randal L. Schwartz - Stonehenge Consulting Services, Inc. - +1 503 777 0095
<merlyn@stonehenge.com> <URL:http://www.stonehenge.com/merlyn/>
Smalltalk/Perl/Unix consulting, Technical writing, Comedy, etc. etc.
See http://methodsandmessages.vox.com/ for Smalltalk and Seaside discussion

Re: "pasteboard doesn't contain valid data" on OSX

[Jan Djärv]

Randal L. Schwartz skrev 2010-08-11 16.53:
>>>>>> "Jan" == Jan Djärv<jan.h.d@swipnet.se>  writes:
>
> Jan>  Works for me with the patch I sent.  But that patch isn't checked in.  You did
> Jan>  recompile and install ns-win.el didn't you?
>
> I'm doing a make bootstrap after checking out HEAD... which seems to
> recompile a lot of files, so I'm guessing yes.
>

That is strange.  I guess we have to wait for the real solution.

	Jan D.

Re: "pasteboard doesn't contain valid data" on OSX

[Chad Brown]

This has been broken since Chong made the `Selection Changes' in the middle of last month.

I haven't dug into it yet because I've been traveling, and because you can work-around the problem by repeating the operation a second time -- i.e. M-w M-w works as M-w should work.  

Hope that helps,
*Chad

Re: "pasteboard doesn't contain valid data" on OSX

[Leo]

On 2010-08-11 19:37 +0100, Chad Brown wrote:
> This has been broken since Chong made the `Selection Changes' in the
> middle of last month.
>
> I haven't dug into it yet because I've been traveling, and because you
> can work-around the problem by repeating the operation a second time
> -- i.e. M-w M-w works as M-w should work.
>
> Hope that helps, *Chad

Sorry to jump into the discussion.

I think I have reported this bug quite a while ago when I was using the
ns-port. The problem is when your pasteboard contains an image and try
to paste it into emacs, you get the error. In mac-port, you can paste
image into a buffer so I have never seen this error again.

Leo

Re: "pasteboard doesn't contain valid data" on OSX

[Randal L. Schwartz]

>>>>> "David" == David De La Harpe Golden <david@harpegolden.net> writes:

David> Seems stuff that's named "cut buffer" in the ns-specific parts
David> has basically nothing to do with x11 cut buffers.  I don't know
David> if there was a misunderstanding, or an active decision to reuse
David> the name, or what (I suppose I could paw through the logs), but
David> it's a terrible name given the opportunity for confusion with an
David> obsolete x11 feature (especially remembering that ns emacs /runs
David> on/ x11 in some incarnations).

By the way, the problem has been cured as of the changes in the past day
or two.  Thanks guys!

-- 
Randal L. Schwartz - Stonehenge Consulting Services, Inc. - +1 503 777 0095
<merlyn@stonehenge.com> <URL:http://www.stonehenge.com/merlyn/>
Smalltalk/Perl/Unix consulting, Technical writing, Comedy, etc. etc.
See http://methodsandmessages.vox.com/ for Smalltalk and Seaside discussion