Eric House
|
7d8f5844bd
|
fix to compile without DEBUG defined
|
2014-03-11 19:38:08 -07:00 |
|
Eric House
|
640837f318
|
up strings for new release
|
2014-03-11 19:16:41 -07:00 |
|
Eric House
|
a3e9562654
|
put related fields together -- no code change
|
2014-03-11 07:02:10 -07:00 |
|
Eric House
|
dee75f5a0a
|
don't ref dicts from dmgr_get()
|
2014-03-10 21:50:06 -07:00 |
|
Eric House
|
a4302897e2
|
enlarge dictmgr to track a compile-time defined number of dicts
|
2014-03-10 21:01:24 -07:00 |
|
Eric House
|
55267f2e42
|
add memmove
|
2014-03-10 20:25:38 -07:00 |
|
Eric House
|
f601f2bf57
|
use dictmgr in java. This requires jni globals that live across boards being opened and that include a mempool dicts will be allocated out of. Seems to work, and to get ref counts as high as 6 or so before they drop down to one when all boards are closed. (The final is held by the dictmgr which won't give it up until the app itself is GC'd)
|
2014-03-10 19:14:59 -07:00 |
|
Eric House
|
3c32ca6402
|
add option to log callsites of dictmgr put and get for easier ref leak tracking
|
2014-03-10 18:59:10 -07:00 |
|
Eric House
|
6203a23c62
|
don't crash in networked games
|
2014-03-10 18:58:16 -07:00 |
|
Eric House
|
fec1ce3bed
|
use pthreads mutex to protect with will become critical sections in dict mgr, as will be accessed from several threads on android
|
2014-03-10 18:56:39 -07:00 |
|
Eric House
|
5500d36cde
|
add and use dict manager that caches, using refcount, dicts across games. Allows multiple games to be opened, and open at the same time, with only one instance of a dict ever created. Works on linux; not used or tested yet on Android.
|
2014-03-09 19:12:21 -07:00 |
|
Eric House
|
a2671f5ef0
|
add global mempool that can survive individual games, and include it every game's globals. Use it instead of game's mempool to create dictionaries so they can outlive a game and not trip asserts on pool delete. Wrap dicts in a java class that refcounts them when BoardCanvas wants to keep one so that, I hope, the bug using a deleted dict will go away.
|
2014-03-09 14:46:33 -07:00 |
|
Eric House
|
9f5d470c07
|
no need for final
|
2014-03-09 14:39:18 -07:00 |
|
Eric House
|
9d9c822bb8
|
save, commented out, how to call the contstructor of a nested class -- in case I go back to doing things that way.
|
2014-03-09 14:23:24 -07:00 |
|
Eric House
|
a7c114e3f9
|
refcount dicts. Model "owns" its copy and so increments the refcount when gaining one and decrements the count on any being replaced (and on all on exit). This is setting up the real change, which is to let the java world wrap dicts in objects that hang onto them until they're destroyed, which should fix problems where dicts are referenced after they've been destroyed.
|
2014-03-08 20:57:27 -08:00 |
|
Eric House
|
8a6adfec34
|
cleanup -- no logic change
|
2014-03-08 20:51:49 -08:00 |
|
Eric House
|
e0c17c50b3
|
catch off-thrown exception. This should prevent a crash, but the fix is probably to figure out why it's happening -- later.
|
2014-03-08 12:55:46 -08:00 |
|
Eric House
|
23b72467bb
|
don't assert unset cookieID, as that's normal (I think) for unconnected case
|
2014-03-08 12:47:58 -08:00 |
|
Eric House
|
fb7c1682b4
|
tweak string
|
2014-03-07 19:10:06 -08:00 |
|
Eric House
|
3b971e0153
|
tweak build timestamp format
|
2014-03-06 22:11:13 -08:00 |
|
Eric House
|
c31c9cb365
|
don't show resend menuitem unless there are moves to resend
|
2014-03-06 07:43:20 -08:00 |
|
Eric House
|
c982fb54d9
|
up strings and notes for new release
|
2014-03-06 06:47:43 -08:00 |
|
Eric House
|
3daa877bee
|
Send from resend asyncTask using factored-out code in CommsTransport. Works for SMS at least.
|
2014-03-05 07:15:34 -08:00 |
|
Eric House
|
923ac6db98
|
show loadDB menuitem only when nothing's selected
|
2014-03-05 06:37:02 -08:00 |
|
Eric House
|
45ace48490
|
tweak string
|
2014-03-05 06:32:03 -08:00 |
|
Eric House
|
bcfd27ee7f
|
don't crash when there's a SMS game with pending messages
|
2014-03-04 21:38:22 -08:00 |
|
Eric House
|
5ebbf87c24
|
don't kick off async task if there's nothing to send
|
2014-03-04 21:13:23 -08:00 |
|
Eric House
|
811be34a20
|
remove logging
|
2014-03-04 21:12:40 -08:00 |
|
Eric House
|
cfee7ef8ac
|
resend all messages from the network change broadcast receiver itself instead of GamesList, which will not be listening if another activity is in front or if the app's in the background.
|
2014-03-04 19:03:39 -08:00 |
|
Eric House
|
3c0b4a899f
|
resend all unacked messages each time we regain network, but only up to four times/hour.
|
2014-03-04 18:36:46 -08:00 |
|
Eric House
|
431b24409f
|
Fix kitkat-only problem where expiring rects are clipped at top and left
|
2014-03-04 07:40:25 -08:00 |
|
Eric House
|
0a601a73a4
|
(commented out) line to cause ndk to be built without optimization for better source-level debugging
|
2014-03-04 07:07:44 -08:00 |
|
Eric House
|
7002ff11c7
|
fix so non-DEBUG build compiles
|
2014-03-04 07:07:01 -08:00 |
|
Eric House
|
df771f50c4
|
add function that resends all pending (un-ackd, at the comms level) messages for unlocked games and, for testing mostly, a menuitem on the main activity to trigger it. Probably crashes if there's a message for a game connected by other than relay.
|
2014-03-04 07:05:55 -08:00 |
|
Eric House
|
60bce8f22b
|
add menuitems recently added to non-small-devices' menu
|
2014-03-02 20:34:58 -08:00 |
|
Eric House
|
9d174740c8
|
don't show "send pending messages" menuitem when there are none
|
2014-03-01 16:15:35 -08:00 |
|
Eric House
|
386c28e596
|
don't notify network change listeners unless a change is in fact a change and until there's been two seconds without changes. There's a lot of connection up-and-down spam when connecting and maybe with iffy connectivity, and no point in passing that through to code that will do stuff like stopping a send thread in response.
|
2014-03-01 15:17:14 -08:00 |
|
Eric House
|
23638310bc
|
register for net state changes, and toast/log on debug builds. This is toward resending from all games with packets pending on reconnect.
|
2014-03-01 14:59:17 -08:00 |
|
Eric House
|
56c05869da
|
allow to run from bin directory, which turns out to be useful
|
2014-03-01 12:39:51 -08:00 |
|
Eric House
|
d122b7d2ff
|
include all .xml and .java files, but only if they're not named in a .gitignore file in the same directory
|
2014-03-01 11:44:20 -08:00 |
|
Eric House
|
c1e22dd3cf
|
add summary DB column and jni code to fetch it to store and display number of unacked messages a game has. And add debug-only option to display it in the main list. In the future I want to be able to iterate over all these games and resend their packets (without opening them visually) when an internet connection comes back up.
|
2014-02-28 18:59:12 -08:00 |
|
Eric House
|
242f722f2e
|
remove logging
|
2014-02-28 18:56:39 -08:00 |
|
Eric House
|
e536a6d3ee
|
format build timestamp added to about dialog
|
2014-02-28 06:47:44 -08:00 |
|
Eric House
|
1d85a74a6c
|
readd old assertions -- and fix compile error
|
2014-02-28 06:47:27 -08:00 |
|
Eric House
|
b45fcf2aa6
|
force output in UTF-8 -- fixes German BYOD display and still works for English
|
2014-02-28 06:27:54 -08:00 |
|
Eric House
|
f42fbf62c4
|
ignore files with names ending in ~
|
2014-02-28 06:16:13 -08:00 |
|
Eric House
|
9a94663fb8
|
log when we drop a message because two possible receiving games are open
|
2014-02-28 06:13:52 -08:00 |
|
Eric House
|
61080e1b62
|
revert changes pausing instead of exiting JNI thread (and so not saving game data) from BoardActivity's onPause() when onFinishing() returned false. This is necessary because there are cases where moves were generated and sent without local state ever being saved because onPause() was called only once on what turned out to be activity termination.
|
2014-02-28 05:57:32 -08:00 |
|
Eric House
|
cc4982b833
|
add debug-only code to assert that movestack is never disposed when dirty (via macros easy to use elsewhere)
|
2014-02-28 05:43:47 -08:00 |
|
Eric House
|
efbb676cce
|
better formatting for hexdump offsets
|
2014-02-28 05:42:22 -08:00 |
|