57ee4d56c8
Let's have some debug logging so that we can investigate potential block resolution wackiness.
Michael Bayne
2003-05-02 18:09:56 +00:00
7c7db998ac
With this fix, and the new JVM, I think full screen actually works now. It's a miracle!
Michael Bayne
2003-05-02 18:03:17 +00:00
02bd2ade00
We just flip flop back and forth on this one: now non-resolved tiles *are* traversable so that if a sprite claims to have relocated somewhere outside your resolved scene bounds, we'll just bee line them to their new location. Since the user can never click outside their resolved view bounds, we won't ever compute an invalid path for them. Joy.
Michael Bayne
2003-05-02 15:13:20 +00:00
e6126a4787
If FPS display is on, dirty it properly when the view scrolls.
Michael Bayne
2003-05-02 15:11:51 +00:00
9454a1e35a
Commented out avg. dirty regions per tick display and modified FPS display to facilitate efficient checking of whether it's activated or not.
Michael Bayne
2003-05-02 15:11:12 +00:00
94697f3e3f
If our sleep granularity is zero, yield rather than call sleep() which would do nothing.
Michael Bayne
2003-05-02 15:09:57 +00:00
31e26c681c
Need to take a different approach to get everything to repaint when we lose our back buffer.
Michael Bayne
2003-05-01 22:06:52 +00:00
b2ba9dc9c6
We'll be needin' to use that width and height that are passed in.
Michael Bayne
2003-05-01 22:04:48 +00:00
098cb693d1
Fixed.
Ray Greenwell
2003-05-01 02:11:22 +00:00
0e8b1dd058
Removed crack pipe.
Ray Greenwell
2003-05-01 02:05:51 +00:00
2c94e12490
Regenerated DObject definitions with new policy.
Michael Bayne
2003-04-30 22:45:57 +00:00
dd2657db24
Update our local variable *after* we request the attribute change so that that code can properly obtain the old value for inclusion with the change/update event.
Michael Bayne
2003-04-30 22:44:48 +00:00
fc6d71cae1
Actually initialize the oldValue to UNSET_OLD_VALUE.
Ray Greenwell
2003-04-30 22:38:49 +00:00
dc15ffc36f
Have to supply the old value now when constructing an AttributeChangedEvent by hand.
Michael Bayne
2003-04-30 22:32:20 +00:00
43edd884c0
Added the necessary machinations to make getOldValue() and getOldEntry() work on the server.
Michael Bayne
2003-04-30 22:32:04 +00:00
99fa7fd6ff
renderMedia() -> paint().
Michael Bayne
2003-04-30 20:48:51 +00:00
357e17d7c7
Because we cleverly do nothing in paint() if there are no dirty regions, we have to go out of our way to let the abstract media managers know when tick() is all done and paint() *may* be called. This way they can properly clear out their tick stamps because they know that no more media fooling around will happen between the call to willPaint() and the call to paint() (which used to be called renderMedia() but is now paint() to remain in line with all the other places were we tick() and then paint()).
Michael Bayne
2003-04-30 20:48:38 +00:00
6cff845a6c
Use our own thread to queue up frame ticks rather than the standard Java Timer which does undesirable things like suffer from 20 ms granularity on Linux and rapid fire multiple invocations of our timer task if it has fallen behind.
Michael Bayne
2003-04-30 06:42:50 +00:00
9372df6ff0
Render the additional performance metrics when such are activated.
Michael Bayne
2003-04-30 06:34:55 +00:00
580fdba83b
Revamped animation observation.
Michael Bayne
2003-04-30 00:45:02 +00:00
99127e4ce0
Revamped sprite observation and added support for being notified if a path is cancelled.
Michael Bayne
2003-04-30 00:44:36 +00:00
3cfccff0c6
The revamp that made me nervous about our rendering performance but turned out to be not at all related.
Michael Bayne
2003-04-30 00:43:54 +00:00
83d7506e0f
Utility class for tracking trailing averages.
Michael Bayne
2003-04-30 00:38:35 +00:00
b2cbd63be2
Display FPS in the media panel where we have tight control over dirtying and repainting.
Michael Bayne
2003-04-29 18:15:10 +00:00
83cdc0b0fb
Canned support for FPS display by the frame manager; it just can't manage the dirty regions effectively.
Michael Bayne
2003-04-29 18:14:37 +00:00
82d3a89e97
Further tweaks though this time they are functionally equivalent.
Michael Bayne
2003-04-28 23:33:43 +00:00
c045a84703
We want to compare the origin of b not its left and right tiles.
Michael Bayne
2003-04-28 22:21:20 +00:00
2a909b772e
When we're resized, delay our rethink until the next tick so that the view bounds are properly updated for the rethink.
Michael Bayne
2003-04-28 21:46:33 +00:00
4d7993b59f
Only delay repaint on setBounds() if we end up queueing up blocks as a result.
Michael Bayne
2003-04-28 21:10:19 +00:00
98d6d2f086
Broke out influential bounds computation into a method that can be overridden.
Michael Bayne
2003-04-28 21:02:54 +00:00
abec16cbc1
Assign an ordering based on y position in the absence of any other ordering.
Michael Bayne
2003-04-28 20:53:58 +00:00
f2c1c7dad5
Updated render sort to reflect arbitrary ordering assigned to non-ordered objects.
Michael Bayne
2003-04-28 20:53:24 +00:00
0165cca20a
Make a note of a potentially non-intuitive return value.
Michael Bayne
2003-04-28 20:41:13 +00:00
da3ec5f6e4
Report unexpected visibility more sensibly.
Michael Bayne
2003-04-28 20:28:45 +00:00
120c2643ef
Added grow() and a Log class.
Michael Bayne
2003-04-28 20:27:52 +00:00
508e389678
Make the bounds of just our objects available as well.
Michael Bayne
2003-04-28 20:27:35 +00:00
8ce7d7bc69
Slap a big fat random number on the end of our cache directory to avoid annoyances when we run multiple clients for testing.
Michael Bayne
2003-04-28 18:21:47 +00:00
7c5d0f275e
Document that little feller.
Michael Bayne
2003-04-28 18:12:09 +00:00
831539b09d
How simple! If our bounds change, delay a repaint until we have all of our blocks resolved. In the future, we could attempt to resolve the blocks that are needed to repaint the visible view first and allow repaint as soon as those blocks are resolved.
Michael Bayne
2003-04-28 18:08:57 +00:00
72768a3a9e
More synchronization is needed! The dangers of multithreaded programming.
Michael Bayne
2003-04-28 17:38:06 +00:00
201ee0acf3
Don't log the whole message, just the class.
Michael Bayne
2003-04-27 07:53:31 +00:00
5c03adbb72
Added support for temporarily disabling image and tile cache flushing.
Michael Bayne
2003-04-27 07:33:15 +00:00
a26d42549f
Enhanced error logging.
Michael Bayne
2003-04-27 06:43:51 +00:00
278b04206c
Test fast loading speed. It's speedy!
Michael Bayne
2003-04-27 06:38:48 +00:00
21889f5a9e
Trimmed tilesets are all fast. This should ensure that it's possible to save the tileset image in fast format before doing so because the set of formats for which we support trimming is slightly larger than those for which we support saving in fast format. We don't use such formats presently, so I'll temporarily punt.
Michael Bayne
2003-04-27 06:38:15 +00:00
57c805d929
Component bundles are all fast.
Michael Bayne
2003-04-27 06:35:28 +00:00
b84f62e35d
If our image has been converted to fast format; load it up fast and propah!
Michael Bayne
2003-04-27 06:35:09 +00:00
cab6cfb131
Convert images to "fast" images if possible. Also fixed a bug where it wasn't cleaning up properly if it failed to generate a bundle. Plus the error reporting is all much better.
Michael Bayne
2003-04-27 06:34:43 +00:00
65106d8815
Generate better error messages if we fail.
Michael Bayne
2003-04-27 06:33:40 +00:00
8973ca0d8c
Make ImageDataProvider responsible for loading the actual image instead of returning an ImageInputStream.
Michael Bayne
2003-04-27 06:33:11 +00:00
a4e09831c5
The amazing fast image loading extravaganza. Images load in 1/4th the time and generate almost zero garbage during the loading process (unlike the ImageIO loaders which practically generate an entire image worth of garbage during the load).
Michael Bayne
2003-04-27 06:32:15 +00:00
a54e01a0fd
Provide estimated travel time.
Michael Bayne
2003-04-27 03:53:41 +00:00
3962a0fa77
Allow our resource cache directory to be configured.
Michael Bayne
2003-04-27 03:27:33 +00:00
e1ba8eeeec
Allow others to know what our cache directory is.
Michael Bayne
2003-04-27 03:27:21 +00:00
e1b4198d8f
Let's just report the type of message.
Michael Bayne
2003-04-27 01:32:08 +00:00
661d7598ed
Commented out verbose logging.
Michael Bayne
2003-04-27 01:30:25 +00:00
055bd6a642
Factored out the flip-buffer versus volatile image back-buffer code into start-time selectable implementations. Modified back buffer to use BufferCapabilities.FlipContents.COPIED which should do what we need in terms of providing us with the current frame's contents on which to paint.
Michael Bayne
2003-04-26 17:56:26 +00:00
b907bc02b0
Ensure that we always eventually claim to be resolved.
Michael Bayne
2003-04-26 02:16:18 +00:00
db1a0f59c9
Don't constrain tooltips to the view bounds because they won't repaint properly and in happy virtual land it's not so necessary any longer.
Michael Bayne
2003-04-26 00:48:47 +00:00
dac610797e
Added a means by which the default base tileset can be set.
Michael Bayne
2003-04-25 22:26:25 +00:00
d86ddce45d
More delayed block loading improvements: resolve fewer blocks, use the same mechanism for including and excluding blocks.
Michael Bayne
2003-04-25 22:13:14 +00:00
8bab0bd281
Added useful debug logging.
Michael Bayne
2003-04-25 22:12:14 +00:00
cca2a9cd16
Beginnings of support for asynchronous scene image data resolution.
Michael Bayne
2003-04-25 18:22:52 +00:00
a03ad10f8d
Beginnings of support for asynchronous resolution of scene image data.
Michael Bayne
2003-04-25 18:22:26 +00:00
2051770c91
Synchronize access to the tile cache so that we can load tiles on other threads.
Michael Bayne
2003-04-25 18:21:49 +00:00
90391aa69d
Synchronize access to the image cache so that we can load images on other threads.
Michael Bayne
2003-04-25 18:21:28 +00:00
40e72feb45
Cleaned up the image loading speed test.
Michael Bayne
2003-04-25 15:52:39 +00:00
a4741de4f8
No more MisoSceneMetrics.bounds.
Michael Bayne
2003-04-25 15:52:25 +00:00
2f80c9b057
Removed obsolete image loaders.
Michael Bayne
2003-04-25 15:51:49 +00:00
ae0376d4ec
Removed a bunch of obsolete stuff.
Michael Bayne
2003-04-25 15:51:34 +00:00
082af9eedd
Add support for putting scene object into warning mode.
Michael Bayne
2003-04-25 01:15:00 +00:00
5d5a0e8173
Allow base tiles to be cleared out; clear out the fringe when a base tile is updated because the caller will need to recompute it.
Michael Bayne
2003-04-24 21:27:23 +00:00
fc9a5dc7e2
Minor restructuring.
Michael Bayne
2003-04-24 21:12:10 +00:00
f894a18bcb
Fixed bug with object removal. Added sanity checking code in addObject() to prevent duplicate objects from being added.
Michael Bayne
2003-04-24 21:11:23 +00:00
26d15d3d23
Corrected documentation: the resource set is supplied, not the specific bundle.
Ray Greenwell
2003-04-24 07:26:45 +00:00
5d6e746b88
Wire up the magic so that data in resource bundles can be acquired via URLs.
Ray Greenwell
2003-04-24 07:23:47 +00:00
ea3adbb530
Use the compose() method which properly handles negative y values.
Michael Bayne
2003-04-23 00:45:24 +00:00
0eea772f57
Oh the delicacy of the negative numbers. Wasn't properly computing neighbor and block keys and was thus booching object footprint coverage computation.
Michael Bayne
2003-04-23 00:37:46 +00:00
3d290efb21
A streamable intint map, it's amazing!
Eric Lundberg
2003-04-22 23:36:56 +00:00
995f33af84
Provide our view bounds to the radial menu services.
Michael Bayne
2003-04-22 22:00:51 +00:00
76cb19646c
Use the default base tileset when computing fringe if an actual base tile is not specified.
Michael Bayne
2003-04-22 20:36:11 +00:00
1274fbb7c0
Let's inform the derived classes of the delta rather than the new _vbounds upper left which is something they're already much more likely to be making use of directly.
Michael Bayne
2003-04-22 18:12:40 +00:00
a555443c04
Log a stack trace to help us find out where this sort of thing happens if it happens.
Michael Bayne
2003-04-22 01:57:44 +00:00
1da61b99c7
Disallow adding the same object in the same place twice.
Michael Bayne
2003-04-22 01:57:20 +00:00
6b0e816643
Dooh! We need to recompute our visible object set regardless of whether or not new scene blocks were resolved.
Michael Bayne
2003-04-21 17:56:55 +00:00
e628a8e429
Added support for specifying a default tileset for scenes.
Michael Bayne
2003-04-21 17:08:57 +00:00
1fc693523d
Add constructor that takes bounds. We love the bounds.
Michael Bayne
2003-04-20 04:52:33 +00:00
0d740682db
We need to be streamable.
Michael Bayne
2003-04-20 00:05:14 +00:00
ac9fc9a35d
Make blockwid and blockhei shorts and make them not be multiples of ten so that we tease out any bugs that may have been hidden by uniform tiling.
Michael Bayne
2003-04-19 23:06:20 +00:00
3c94cd67f6
Various fixes to support negative tile coordinates; created sparse miso scene model and associated business to store scene data in an arbitrary set of blocks rather than fixed arrays; finally stuck a fork in scene width and height as well as view width and height along with origin offset and a few other now obsolete bits.
Michael Bayne
2003-04-19 22:40:34 +00:00
95b19feb51
The scene metrics are now *much* simpler.
Michael Bayne
2003-04-19 22:37:57 +00:00
1cb3d1edb5
Added floorDiv().
Michael Bayne
2003-04-19 22:28:28 +00:00
55ecaf885a
Out with the obsolete. The "oh, so" obsolete.
Michael Bayne
2003-04-19 20:44:10 +00:00
299e0845bb
Center on the center of the scene when it is first set. Adjust our view bounds to preserve the current center point when we are resized. Deep sixed "bounded" and some other needless fiddling.
Michael Bayne
2003-04-19 17:06:08 +00:00
bdd1697a41
Don't repaint when we rethink, rearranged some things so that we can avoid recomputing visible objects, etc. if no blocks were loaded or flushed.
Michael Bayne
2003-04-19 02:01:56 +00:00
7772cdc175
Make our bounds available.
Michael Bayne
2003-04-19 02:01:08 +00:00
626d59a718
Apparently revalidating ain't going to do it, so we recenter and rethink.
Michael Bayne
2003-04-19 01:04:57 +00:00
e6a1ebbd80
We need to recreate our back buffer if our frame's bounds change.
Michael Bayne
2003-04-19 01:04:29 +00:00
3abbd1c8af
Removed old testing code. Whoopskidoodle!
Michael Bayne
2003-04-19 00:56:35 +00:00
Michael Bayne