Commit graph

225 commits

Author SHA1 Message Date
Eric House
a34a859d18 log channelNo more consistently 2014-11-06 18:44:19 -08:00
Eric House
4d2f49d33f when sending and don't have an address for channel yet, send the one
comms was created with, which is what client will have to do anyway.
2014-09-14 13:41:12 -07:00
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
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
025bcb885b don't move state forward when should be moving back 2014-02-09 19:35:17 -08:00
Eric House
0b132661ca change struct name from r to rr for easier searching; no code change 2014-02-07 21:12:45 -08:00
Eric House
e34cc9007d fix problems with opening and using wordlists on 64-bit linux by
making XP_U32 32 rather than 64 bits. Duh. Do same on Android jni just
in case.
2014-01-07 06:58:20 -08:00
Eric House
de41802f54 rename variable so I don't trip up again 2013-12-17 21:08:27 -08:00
Eric House
4420049bd0 remove duplicate lines (added by merge?) 2013-12-16 08:05:35 -08:00
Eric House
1f877c7567 remove logging 2013-07-29 07:29:01 -07:00
Eric House
a191c6fc2a don't add chats to queue at len 64 rather than 128 2013-07-22 06:12:06 -07:00
Eric House
27d9f3de45 Don't enable chat when more than 128 messages are in queue to prevent
an extremely unlikely cause of stalls -- and make tests trip over full
queues less often.
2013-07-18 07:37:45 -07:00
Eric House
62dacb83d8 fix to compile when DEBUG not defined 2013-07-16 08:17:46 -07:00
Eric House
9e358ad413 don't insert a new message in queue if it's identical to the one
currently at the end.  This mostly happens when devices are making
several initial client connection attempts before succeeding.
2013-07-12 07:34:12 -07:00
Eric House
a522266ce4 add message checksumming for android too by making it part of util
(compiled out by default)
2013-07-10 08:05:47 -07:00
Eric House
5462a05692 Merge branch 'gtk_multigame' into android_branch 2013-07-09 19:35:26 -07:00
Eric House
bf1cdf225c print queue less frequently 2013-07-09 19:32:49 -07:00
Eric House
d50c808f96 Merge branch 'android_branch' into gtk_multigame
Conflicts:
	xwords4/android/XWords4/src/org/eehouse/android/xw4/DlgDelegate.java
	xwords4/common/comms.c
	xwords4/linux/cursesmain.c
	xwords4/linux/cursesmain.h
	xwords4/linux/gtkmain.c
	xwords4/linux/gtkmain.h
	xwords4/linux/linuxmain.c
	xwords4/linux/main.h
	xwords4/linux/scripts/discon_ok2.sh
	xwords4/relay/xwrelay.cpp

(Note: The curses app crashes on exit with mempool assertions, but that's a problem before the merge.)
2013-07-09 07:18:00 -07:00
Eric House
a3de759d9e fix to compile on Android when DEBUG set 2013-07-07 19:40:07 -07:00
Eric House
923d41e0ee fix to compile when DEBUG not set 2013-07-07 13:17:26 -07:00
Eric House
bbac923f68 assert queueLen at the point where it'll be easier track why it's overgrowing 2013-07-05 21:32:41 -07:00
Eric House
7229535be7 insist on non-0 channel ID *net of mask*; don't send acks when we
don't have a connID yet, as this confuses the server.  This fixes
assertions in testing that were pretty common when chat was enabled;
it may also fix the reported problem on Android that games get hosed
when people are using chat, as the server's response (without the
assert) was to add a new channel.
2013-07-03 07:39:05 -07:00
Eric House
858d00332a when an initial message from a client arrives with channel already
set, drop it (after asserting in debug code.)  I'm not yet sure why
this happens, but giving it a new channel is not the right move.
2013-07-02 18:49:48 -07:00
Eric House
1f63745e34 add more logging in search of channelNo bug 2013-07-01 07:54:38 -07:00
Eric House
c77021231c print checksums of sent and received messages 2013-06-30 08:33:29 -07:00
Eric House
b31858ee5e log rather than assert when hostid changes -- though I'm not seeing
this now that the relay doesn't send packets to the wrong device
2013-06-26 00:42:55 -07:00
Eric House
6d22c3ab86 tweak logging 2013-06-22 05:51:29 -07:00
Eric House
8e8b4755b3 bring in changes from gtk_multigame to remove assertion fail when
connname gets changed by relay due to failure to ack it (which happens
when a large number of games is being run in a test.)
2013-05-30 06:48:29 -07:00
Eric House
d78a4979da HACK: if gi's notion of gameID is out of sync, force it in (logging
the fact).  I'm not sure why but this is required sometimes on
Android.
2013-01-28 07:01:49 -08:00
Eric House
3be45a67f7 fix erroneous asserts: it's uncommon but ok for a device to get
assigne a second connname.  (Cause: failure of relay to receive the
device's ACK.)
2013-01-26 18:56:53 -08:00
Eric House
c839db4a25 disconnect from relay no matter what the state 2013-01-22 06:12:55 -08:00
Eric House
f6e7ece578 add comms_stop() and an assert to make sure it's called. Explicit
stop and a message to the relay when using a UDP socket makes up for
there being no notification of connection closing.
2013-01-16 06:31:09 -08:00
Eric House
5463535265 don't increase backoff if there wasn't actually anything to send. 2012-11-14 05:42:47 -08:00
Eric House
c4df31892e Fix occasional infinite loop of devices sending and receiving the same
message over and over when getting updated by GCM.  The problem
occurred when one device had an un-ACKable initial message still in
its queue.  I call resendAll a lot, which caused that message to get
resent to the other game which then replied without being able to ACK
it so it remained to be sent again.  This would continue until users
moved forward in the game.  The fix is to add a backoff timer to
resendAll() so that it can't loop.  The timer is reset when an ackable
and new message is received, meaning there's been a change in what's
available to resend.  And since users calling resendAll manually
expect it to do something, add a force param that ignores the backoff.
seems to fix the problem (but needs a lot of testing.)
2012-11-13 21:25:03 -08:00
Eric House
cf8d30ce5d tweak logging; fix uninitialized debug-only variable 2012-11-13 21:18:52 -08:00
Eric House
57ea768aaf refuse to send 0-length messages. That may happen only when the stupid
curses client calls server_initClientConnection over and over, but
reduces unnecessary messaging in that case at least.
2012-11-13 07:22:48 -08:00
Eric House
43b1e219f3 add a bit more logging to comms 2012-11-13 06:14:36 -08:00
Eric House
1316ae4b67 Fix relay devid protocol to deal with case where client submits a
ID_TYPE_RELAY id that's not in the devices table (as has happened when
a device switches relay URLs during testing, but might also happen if
I have to delete an entry from the devices table.)  In that case,
return ID_TYPE_NONE to the client, which will be its clue to delete
its ID_TYPE_RELAY id and submit the platform-specific id again.

Note: android won't compile this revision thanks to util.h change
2012-11-10 15:05:44 -08:00
Eric House
f1ee77882b Send devid with RECONN as well as CONN so games present when device
upgrades can also use GCM.  Tested on relay but not device.
2012-11-09 06:54:12 -08:00
Eric House
9ad28908bb assert enum within range 2012-11-06 07:07:14 -08:00
Eric House
59937b8514 modify how devids are handled on the relay and the protocol through
which they're communicated to the device.  Device is expected to have
a platform-specific notion of ID which the relay stores in a new
devices table and indexes with a 32-bit number which is returned to
the device -- which is encouraged but not required to use it in lieu
of the longer ID in future communications.  Modify linux client and
test script to use the relay-supplied id.  Some of this is commented
out for now.
2012-11-03 10:58:01 -07:00
Eric House
3469975d2a add, still disabled by compile-time flags, code to transmit device ID
(e.g. ID Android devices get via GCM) to relay and to associate it
there with messages that need to be delivered.
2012-10-30 07:01:47 -07:00
Eric House
6ddd4a61c1 fix compile errors (mostly unused variables) found by gcc on latest
Ubuntu (12.04).
2012-09-25 07:28:02 -07:00
Eric House
0ccc2331b5 address bug where message was getting ACK'd by a device that failed to
save what it had ACK'd leaving the game permanently broken.  Do that
by adding a new method game_saveSucceeded() called after the client
claims to have committed bytes returned by game_writeToStream() to
disk.  In that method comms updates the value it'll use in subseqent
ACKs.
2012-09-10 07:31:45 -07:00
Eric House
32e1530d3f i => ii; no other change 2012-08-22 07:44:21 -07:00
Eric House
08f03af2cc don't write gameID as 32 bits when it's bound for old-version client,
and assert based only on the lower 16 bits in case came from
old-version client.  Fixes crashes testing mid-game upgrade from beta
43.
2012-05-07 07:02:43 -07:00
Eric House
033a69be8d to make it easier to tell games apart, print the number[s] of the
remote device[s] as part of summary view and in game config screen
(read-only).  Use same field in summaries table for remote phone
numbers and bt addresses.
2012-04-18 23:23:36 -07:00
Eric House
416a1054a7 Merge branch 'android_ack' into android_bt
Conflicts:
	xwords4/common/comms.c
	xwords4/common/comtypes.h
2012-04-13 20:57:35 -07:00
Eric House
f22d9530b1 track what's been ack'd, and add method to ack what hasn't been. 2012-04-13 20:21:59 -07:00
Eric House
404265565b tighten definition of connected BT or SMS game 2012-04-12 22:14:13 -07:00