Re: [Emacs.app dev]: ghost cursor problem is still there
On 23 Aug 2008, at 13:30, Dan Nicolaescu wrote: + drawGlyph = 1; // just draw the Glyph [FRAME_BACKGROUND_COLOR (f) set]; ! ! NSDisableScreenUpdates (); Unfortunately this breaks GNUStep: Not a problem; I have checked in a fix. Can you let us know how well this works on GNUStep? Is there a lot of flicker when moving the cursor or in blink-cursor- mode? Also please make sure that your log entries are properly formatted: Like ChangeLog, it should contain the changed function name, OK, I can do that. I modeled my earlier comments after others in that file, e.g. from 2008-08-01. Could you explain to me how the ChangeLog and the CVS comments should differ semantically? smime.p7s Description: S/MIME cryptographic signature - This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100url=/___ Emacs-app-dev- mailing list Emacs-app-dev-@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emacs-app-dev-
Re: [Emacs.app dev]: ghost cursor problem is still there
On Aug 21, 2008, at 5:08 AM, David Reitter wrote: On 21 Aug 2008, at 03:33, Nick Roberts wrote: There were problems with this patch, e.g., ns-cursor-blink-rate not defined which appears to have been corrected Thanks Adrian for catching the missing checkin. but I still get grey strips which are opaque to the text until the cursor moves over them or I select the region with the mouse. Can you reproduce starting with Emacs -Q? I can. After startup, select a row of text, click onto another application, then back. I don't know if this was there before the patch. As far as the (possibly separate) ghost cursor problem, I narrowed down that it appeared between rc2a and rc3. I'm studying the source code diffs to determine the problem. Adrian - This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100url=/ ___ Emacs-app-dev- mailing list Emacs-app-dev-@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emacs-app-dev-
Re: [Emacs.app dev]: ghost cursor problem is still there
On Aug 21, 2008, at 8:07 AM, Adrian Robert wrote: On Aug 21, 2008, at 5:08 AM, David Reitter wrote: On 21 Aug 2008, at 03:33, Nick Roberts wrote: but I still get grey strips which are opaque to the text until the cursor moves over them or I select the region with the mouse. Can you reproduce starting with Emacs -Q? I can. After startup, select a row of text, click onto another application, then back. I don't know if this was there before the patch. I checked, it was not. In addition the patch introduces the following bug: the cursor is not changed to hollow box from block when focus on the frame is taken away. And when multiple frames are present, things get very messy. If these problems cannot be resolved quickly we should probably revert the patch, or at least the non blink-related parts of it. - This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100url=/ ___ Emacs-app-dev- mailing list Emacs-app-dev-@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emacs-app-dev-
Re: [Emacs.app dev]: ghost cursor problem is still there
On 21 Aug 2008, at 12:07, Adrian Robert wrote: As stated before on the list, the defaults are a replacement for X resources. The latter ARE read on X systems when -q/Q is given. I think the reason the behavior is unexpected under NS is that unlike other systems, NS emacs will WRITE into the defaults database (to save user options). The attached patch will therefore cause these resources not to be read, only under NS, if -q/Q is given. Why save the preferences at all? There is a strong group of people who want Emacs to behave in the same way across systems. Would it really be a good idea if the base Emacs started, from version 23, saving the options at the end, especially since they don't go into `custom-file'. D smime.p7s Description: S/MIME cryptographic signature - This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100url=/___ Emacs-app-dev- mailing list Emacs-app-dev-@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emacs-app-dev-
Re: [Emacs.app dev]: ghost cursor problem is still there
On Aug 21, 2008, at 12:27 PM, David Reitter wrote: On 21 Aug 2008, at 12:07, Adrian Robert wrote: As stated before on the list, the defaults are a replacement for X resources. The latter ARE read on X systems when -q/Q is given. I think the reason the behavior is unexpected under NS is that unlike other systems, NS emacs will WRITE into the defaults database (to save user options). The attached patch will therefore cause these resources not to be read, only under NS, if -q/Q is given. Why save the preferences at all? There is a strong group of people who want Emacs to behave in the same way across systems. Would it really be a good idea if the base Emacs started, from version 23, saving the options at the end, especially since they don't go into `custom-file'. I'd rather energies were focused on fixing bugs for now. But if the defaults are disruptive to that endeavor, what would be the easiest way to stop the problem? The patch I posted is one way. Another way would be to switch ns-save-options to use the customize system (customize-save-variable), which should write it into the .emacs file. The defaults become read-only like X resources, and the prefs window can be removed later with minimal disruption, since the features would be accessible through customize. One would need to be careful with the face-related attributes here, I suspect. - This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100url=/ ___ Emacs-app-dev- mailing list Emacs-app-dev-@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emacs-app-dev-
Re: [Emacs.app dev]: ghost cursor problem is still there
On Aug 20, 2008, at 1:22 AM, David Reitter wrote: Adrian et al, On 22 Jul 2008, at 09:41, David Reitter wrote: The main changes are that we check cursor_type instead of cursorType and draw the text glyph rather than the cursor when erasing anything (`hl' variable). There's a range of steps that we do to ensure that the cursor area is actually visible; I'm not sure if they are really needed, but the corresponding X code does it, too. There is a good bit of guess-work involved, but I'm sure that testing will take care of any problems. OK, in my latest build, things work a lot better and the original blink-cursor-mode appears to work fine now. (I don't quite understand which change made the event mechanisms work better.) This is great news. To repeat, these changes address the following issues: - frame background rather than the right glyph in the white-out phases during blinking - `cursor-type' variable as in core Emacs, rather than NS specific solution - with it, support for things like (box . 2) - box/hollow cursors too narrow OK, these were from before. - standard blink-cursor-mode with all its bells and whistles (whether one needs them or not) This is new, and good. I would also take the blink rate stuff out of the preferences (a patch to the nib) - it doesn't work with the generic blinking code and I believe it's there for historic reasons (because the NS port implemented blinking separately) rather than because it would be very important to have for users (one could think of much more important settings that could be there). This is a matter of opinion and at least I personally find blink and other cursor attributes I like to change easily and frequently. It should be an extremely simple matter to update the prefs controller code to set the core emacs blink rate, and I will do so once you've checked your changes in. The prefs window may be removed completely before 23.1 as it faces hostility from many quarters so I'd rather keep it fully functional until then. The occasional ghost cursors seem to remain. I've been occasionally looking at this when I've had the chance. It has not been easy to track down. I get it (Leopard) in latest source and for a while back to around the merge, but not in rc2a. I think the best chance will be to try to find the first date/version when the problem appeared and then diff the source code (xdisp.c, dispnew.c, and xterm.c will probably tell the story). If anyone has information on this, please post it. Any objections? I'm fine with this being committed. thanks, Adrian - This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100url=/ ___ Emacs-app-dev- mailing list Emacs-app-dev-@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emacs-app-dev-
Re: [Emacs.app dev]: ghost cursor problem is still there
On 20 Aug 2008, at 08:27, Adrian Robert wrote: I'm fine with this being committed. done. - D smime.p7s Description: S/MIME cryptographic signature - This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100url=/___ Emacs-app-dev- mailing list Emacs-app-dev-@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emacs-app-dev-
Re: [Emacs.app dev]: ghost cursor problem is still there
Adrian et al,
On 22 Jul 2008, at 09:41, David Reitter wrote:
The main changes are that we check cursor_type instead of cursorType
and draw the text glyph rather than the cursor when erasing anything
(`hl' variable). There's a range of steps that we do to ensure that
the cursor area is actually visible; I'm not sure if they are really
needed, but the corresponding X code does it, too.
There is a good bit of guess-work involved, but I'm sure that
testing will take care of any problems.
OK, in my latest build, things work a lot better and the original
blink-cursor-mode appears to work fine now. (I don't quite understand
which change made the event mechanisms work better.)
To repeat, these changes address the following issues:
- frame background rather than the right glyph in the white-out phases
during blinking
- `cursor-type' variable as in core Emacs, rather than NS specific
solution
- with it, support for things like (box . 2)
- box/hollow cursors too narrow
- standard blink-cursor-mode with all its bells and whistles (whether
one needs them or not)
I would also take the blink rate stuff out of the preferences (a patch
to the nib) - it doesn't work with the generic blinking code and I
believe it's there for historic reasons (because the NS port
implemented blinking separately) rather than because it would be very
important to have for users (one could think of much more important
settings that could be there).
ns_set_cursor_type is now equivalent to the x_set_cursor_type (quasi)
generic. I left ns_set_cursor_color for you or someone else.
The occasional ghost cursors seem to remain.
Any objections?
D
Index: nsterm.m
===
RCS file: /sources/emacs/emacs/src/nsterm.m,v
retrieving revision 1.23
diff -c -r1.23 nsterm.m
*** nsterm.m5 Aug 2008 03:05:14 - 1.23
--- nsterm.m20 Aug 2008 05:18:41 -
***
*** 168,180
the Function modifer (laptops). May be any of the modifier lisp
symbols. */
Lisp_Object ns_function_modifier;
- /* A floating point value specifying the rate at which to blink the
cursor.
-YES indicates 0.5, NO indicates no blinking. */
- Lisp_Object ns_cursor_blink_rate;
-
- /* Used for liason with core emacs cursor-blink-mode. */
- Lisp_Object ns_cursor_blink_mode;
-
/* A floating point value specifying vertical stretch (positive) or
shrink
(negative) of text line spacing. Zero means default spacing.
YES indicates 0.5, NO indicates 0.0. */
--- 168,173
***
*** 235,241
static NSEvent *last_appdefined_event = 0;
static NSTimer *timed_entry = 0;
static NSTimer *fd_entry = nil;
- static NSTimer *cursor_blink_entry = nil;
static NSTimer *scroll_repeat_entry = nil;
static fd_set select_readfds, t_readfds;
static struct timeval select_timeout;
--- 228,233
***
*** 2270,2275
--- 2262,2271
int on_p, int active_p)
/*
--
External call (RIF): draw cursor
+ (modeled after x_draw_window_cursor and erase_phys_cursor.
+ FIXME: erase_phys_cursor is called from display_and_set_cursor,
+ called from update_window_cursor/x_update_window_end/...
+ Why do we have to duplicate this code?
-- */
{
NSRect r, s;
***
*** 2278,2291
struct glyph *phys_cursor_glyph;
int overspill;
unsigned char drawGlyph = 0, cursorType, oldCursorType;
NSTRACE (dumpcursor);
! if (!on_p)
return;
w-phys_cursor_type = cursor_type;
! w-phys_cursor_on_p = 1;
if (cursor_type == NO_CURSOR)
{
--- 2274,2296
struct glyph *phys_cursor_glyph;
int overspill;
unsigned char drawGlyph = 0, cursorType, oldCursorType;
+ int new_cursor_type;
+ int new_cursor_width;
+ int active_cursor;
+ enum draw_glyphs_face hl;
+ struct glyph_matrix *active_glyphs = w-current_matrix;
+ Display_Info *dpyinfo = FRAME_X_DISPLAY_INFO (f);
+ int hpos = w-phys_cursor.hpos;
+ int vpos = w-phys_cursor.vpos;
+ struct glyph_row *cursor_row;
NSTRACE (dumpcursor);
! if (!on_p) // check this? !w-phys_cursor_on_p)
return;
w-phys_cursor_type = cursor_type;
! w-phys_cursor_on_p = on_p;
if (cursor_type == NO_CURSOR)
{
***
*** 2318,2326
if (overspill 0)
r.size.width -= overspill;
- /* TODO: 23: use emacs stored f-cursor_type instead of ns-
specific */
oldCursorType = FRAME_CURSOR (f);
cursorType = FRAME_CURSOR (f) = FRAME_NEW_CURSOR (f);
f-output_data.ns-current_cursor_color
= f-output_data.ns-desired_cursor_color;
--- 2323,2332
if (overspill 0)
r.size.width -= overspill;
oldCursorType = FRAME_CURSOR (f);
cursorType = FRAME_CURSOR (f) =
Re: [Emacs.app dev]: ghost cursor problem is still there
On Jul 21, 2008, at 11:02 PM, David Reitter wrote: On 19 Jul 2008, at 19:16, Adrian Robert wrote: It's been a longstanding TODO to normalize the cursor handling under NS. Both for storing the shape, as mentioned, and also the blink handling. Most of the code that's in there dates from before emacs itself had these features, and since it has been mostly working it's been less pain to keep it working than to rip it out and start afresh with the emacs stuff. Time for that to change, and I'd definitely welcome help or a patch for all or part of it. The code below takes care of - frame background rather than the right glyph in the white-out phases during blinking - `cursor-type' variable as in core Emacs, rather than NS specific solution - with it, support for things like (box . 2) - box/hollow cursors too narrow Thanks, great stuff! But could you please summarize the changes? It shows like every line was changed (maybe because of indentation?) but it doesn't look like that was actually the case. Could you regenerate the patch w/o indentation/tab changes? I don't understand why internal-show-cursor and the w-cursor_off_p (that it sets) do not work in the NS port - they are checked by redisplay_internal. Is there any commentary regarding stuff like that? This is the way it is.. ;-/ Usually you just have to get your hands dirty and try to trace it back -- where does w-cursor_off_p get set, when/how are other terms calling or triggering functions that set it or trigger calling of draw_window_cursor(), etc.. I mean, the original authors of this code went to lengths to replace the blink-cursor-mode with something off their own (which is a lot less configurable). As I said before, blinking cursor did not exist in core emacs when the NS port added this functionality. (Along with different cursor shapes.) Did you have any sort of a start on using the emacs common code (even though it didn't work)? Maybe if you post a patch for that others could take a stab at moving it further. - This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100url=/ ___ Emacs-app-dev- mailing list Emacs-app-dev-@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emacs-app-dev-
Re: [Emacs.app dev]: ghost cursor problem is still there
On 22 Jul 2008, at 08:39, Adrian Robert wrote:
Thanks, great stuff! But could you please summarize the changes?
It shows like every line was changed (maybe because of indentation?)
but it doesn't look like that was actually the case. Could you
regenerate the patch w/o indentation/tab changes?
Okay, have a look at what's below.
The main changes are that we check cursor_type instead of cursorType
and draw the text glyph rather than the cursor when erasing anything
(`hl' variable). There's a range of steps that we do to ensure that
the cursor area is actually visible; I'm not sure if they are really
needed, but the corresponding X code does it, too.
There is a good bit of guess-work involved, but I'm sure that testing
will take care of any problems.
I don't understand why internal-show-cursor and the w-
cursor_off_p (that it sets) do not work in the NS port - they are
checked by redisplay_internal. Is there any commentary regarding
stuff like that?
This is the way it is.. ;-/ Usually you just have to get your
hands dirty and try to trace it back -- where does w-cursor_off_p
get set, when/how are other terms calling or triggering functions
that set it or trigger calling of draw_window_cursor(), etc..
Yeah, that's how I've been working... it doesn't really feel like
2008. (And not even like 1999.)
Did you have any sort of a start on using the emacs common code
(even though it didn't work)? Maybe if you post a patch for that
others could take a stab at moving it further.
Well, if someone could explain why redisplay noticed a change in -
cursor_off_p in Carbon or X, but not in NS, then that would already
be helpful. I don't think enabling the original blink-cursor-mode
will be difficult, and you get a lot of functionality for free that way.
I'm trying again with a few extra TRACE steps enabled.
cc'ing emacs-devel for some advice regarding the patch.
- D
===
RCS file: /sources/emacs/emacs/src/nsterm.m,v
retrieving revision 1.11
diff -c -w -r1.11 nsterm.m
*** nsterm.m21 Jul 2008 17:47:24 - 1.11
--- nsterm.m22 Jul 2008 13:30:21 -
***
*** 2274,2279
--- 2274,2285
int on_p, int active_p)
/*
--
External call (RIF): draw cursor
+ (modeled after x_draw_window_cursor and erase_phys_cursor.
+ FIXME: erase_phys_cursor is called from display_and_set_cursor,
+ called from update_window_cursor/x_update_window_end/...
+ Why do we have to duplicate this code?
+ Also, why doesn't cursor_off_p (internal_show_cursor) work?
+ This prevents the original blink-cursor-mode from working.
-- */
{
NSRect r, s;
***
*** 2282,2295
struct glyph *phys_cursor_glyph;
int overspill;
unsigned char drawGlyph = 0, cursorType, oldCursorType;
NSTRACE (dumpcursor);
! if (!on_p)
return;
w-phys_cursor_type = cursor_type;
! w-phys_cursor_on_p = 1;
if (cursor_type == NO_CURSOR)
{
--- 2288,2310
struct glyph *phys_cursor_glyph;
int overspill;
unsigned char drawGlyph = 0, cursorType, oldCursorType;
+ int new_cursor_type;
+ int new_cursor_width;
+ int active_cursor;
+ enum draw_glyphs_face hl;
+ struct glyph_matrix *active_glyphs = w-current_matrix;
+ Display_Info *dpyinfo = FRAME_X_DISPLAY_INFO (f);
+ int hpos = w-phys_cursor.hpos;
+ int vpos = w-phys_cursor.vpos;
+ struct glyph_row *cursor_row;
NSTRACE (dumpcursor);
! if (!on_p) // check this? !w-phys_cursor_on_p)
return;
w-phys_cursor_type = cursor_type;
! w-phys_cursor_on_p = on_p;
if (cursor_type == NO_CURSOR)
{
***
*** 2322,2333
--- 2337,2350
if (overspill 0)
r.size.width -= overspill;
+
/* PENDING: 23: use emacs stored f-cursor_type instead of ns-
specific */
oldCursorType = FRAME_CURSOR (f);
cursorType = FRAME_CURSOR (f) = FRAME_NEW_CURSOR (f);
f-output_data.ns-current_cursor_color
= f-output_data.ns-desired_cursor_color;
+
/* PENDING: only needed in rare cases with last-resort font in
HELLO..
should we do this more efficiently? */
ns_clip_to_row (w, glyph_row, -1, NULL);
***
*** 2346,2357
if (cursorType == no_highlight || cursor_type == NO_CURSOR)
{
/* clearing for blink: erase the cursor itself */
[FRAME_BACKGROUND_COLOR (f) set];
! cursorType = oldCursorType; /* just clear what we had before */
}
else
[FRAME_CURSOR_COLOR (f) set];
if (!active_p)
{
/* inactive window: ignore what we just set and use a hollow
box */
--- 2363,2441
if (cursorType == no_highlight || cursor_type == NO_CURSOR)
{
Re: [Emacs.app dev]: ghost cursor problem is still there
On 19 Jul 2008, at 19:16, Adrian Robert wrote:
It's been a longstanding TODO to normalize the cursor handling under
NS. Both for storing the shape, as mentioned, and also the blink
handling. Most of the code that's in there dates from before emacs
itself had these features, and since it has been mostly working it's
been less pain to keep it working than to rip it out and start
afresh with the emacs stuff. Time for that to change, and I'd
definitely welcome help or a patch for all or part of it.
The code below takes care of
- frame background rather than the right glyph in the white-out phases
during blinking
- `cursor-type' variable as in core Emacs, rather than NS specific
solution
- with it, support for things like (box . 2)
- box/hollow cursors too narrow
The code clearly needs further revision (see FIXME), but I don't know
what to do. If you do, please look into it, otherwise I suggest we
take this as iterative improvement and get back to it after more
important things have been dealt with. The whole redisplay stuff
appears complicated and not very well documented.
Apart from this: the way the blinking cursor is implemented is half-
hearted at beast (or rather, a hack).
With this, the cursor blinks even if it is being moved around, which
is really not good. I wonder why no-one has complained about this
yet...
I don't understand why internal-show-cursor and the w-cursor_off_p
(that it sets) do not work in the NS port - they are checked by
redisplay_internal. Is there any commentary regarding stuff like that?
I mean, the original authors of this code went to lengths to replace
the blink-cursor-mode with something off their own (which is a lot
less configurable).
*** nsterm.m21 Jul 2008 20:32:30 -0400 1.11
--- nsterm.m21 Jul 2008 22:57:54 -0400
***
*** 2274,2279
--- 2274,2285
int on_p, int active_p)
/*
--
External call (RIF): draw cursor
+ (modeled after x_draw_window_cursor and erase_phys_cursor.
+ FIXME: erase_phys_cursor is called from display_and_set_cursor,
+ called from update_window_cursor/x_update_window_end/...
+ Why do we have to duplicate this code?
+ Also, why doesn't cursor_off_p (internal_show_cursor) work?
+ This prevents the original blink-cursor-mode from working.
-- */
{
NSRect r, s;
***
*** 2282,2295
struct glyph *phys_cursor_glyph;
int overspill;
unsigned char drawGlyph = 0, cursorType, oldCursorType;
NSTRACE (dumpcursor);
! if (!on_p)
return;
w-phys_cursor_type = cursor_type;
! w-phys_cursor_on_p = 1;
if (cursor_type == NO_CURSOR)
{
--- 2288,2310
struct glyph *phys_cursor_glyph;
int overspill;
unsigned char drawGlyph = 0, cursorType, oldCursorType;
+ int new_cursor_type;
+ int new_cursor_width;
+ int active_cursor;
+ enum draw_glyphs_face hl;
+ struct glyph_matrix *active_glyphs = w-current_matrix;
+ Display_Info *dpyinfo = FRAME_X_DISPLAY_INFO (f);
+ int hpos = w-phys_cursor.hpos;
+ int vpos = w-phys_cursor.vpos;
+ struct glyph_row *cursor_row;
NSTRACE (dumpcursor);
! if (!on_p) // check this? !w-phys_cursor_on_p)
return;
w-phys_cursor_type = cursor_type;
! w-phys_cursor_on_p = on_p;
if (cursor_type == NO_CURSOR)
{
***
*** 2322,2333
--- 2337,2350
if (overspill 0)
r.size.width -= overspill;
+
/* PENDING: 23: use emacs stored f-cursor_type instead of ns-
specific */
oldCursorType = FRAME_CURSOR (f);
cursorType = FRAME_CURSOR (f) = FRAME_NEW_CURSOR (f);
f-output_data.ns-current_cursor_color
= f-output_data.ns-desired_cursor_color;
+
/* PENDING: only needed in rare cases with last-resort font in
HELLO..
should we do this more efficiently? */
ns_clip_to_row (w, glyph_row, -1, NULL);
***
*** 2346,2396
if (cursorType == no_highlight || cursor_type == NO_CURSOR)
{
/* clearing for blink: erase the cursor itself */
[FRAME_BACKGROUND_COLOR (f) set];
- cursorType = oldCursorType; /* just clear what we had before */
}
else
[FRAME_CURSOR_COLOR (f) set];
! if (!active_p)
! {
! /* inactive window: ignore what we just set and use a hollow
box */
! cursorType = hollow_box;
! [FRAME_CURSOR_COLOR (f) set];
! }
! switch (cursorType)
! {
! case no_highlight:
! break;
! case filled_box:
! NSRectFill (r);
! drawGlyph = 1;
! break;
! case hollow_box:
! NSRectFill (r);
! [FRAME_BACKGROUND_COLOR (f) set];
! NSRectFill (NSInsetRect (r, 1, 1));
! [FRAME_CURSOR_COLOR (f) set];
! drawGlyph = 1;
! break;
!
Re: [Emacs.app dev]: ghost cursor problem is still there
Is there a specific reason why a lot of NS specific code is used to store the cursor as a frame parameter? There's this wonderful `get_window_cursor_type' from xdisp.c that could be used to get the right cursor type, which also allows stuff like (bar . 2), which isn't supported right now. It's been a longstanding TODO to normalize the cursor handling under NS. Both for storing the shape, as mentioned, and also the blink handling. Most of the code that's in there dates from before emacs itself had these features, and since it has been mostly working it's been less pain to keep it working than to rip it out and start afresh with the emacs stuff. Time for that to change, and I'd definitely welcome help or a patch for all or part of it. -Adrian - This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100url=/ ___ Emacs-app-dev- mailing list Emacs-app-dev-@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emacs-app-dev-
