Lots of work dealing with banned permissions (SEND_SMS and
RECEIVE_SMS). First, if they're banned and NBSProxy is installed, just
use it as if the permissions had been granted. When it's not there,
explain at various points where users will otherwise be confused: when
they try to invite using data sms, or when they open a game that already
uses it.)
It's an error to use an env with the wrong thread, but it IS possible
for more than one thread to be associated with the same env. So don't
assert to the contrary.
Seeing the occasional crash and trying to plug leaks that might be
contributing. I think I fixed the Gameptr.finalize() assertions I've
been seeing forever but the thread match assertion remains. Logs will
help catch that.
Didn't think it was required, but once the app started getting messages
in the background it stoped working or crashed. Changes are minimal
because this service isn't responsible for receiving messages: it
doesn't have the long-lived threads of the BT and Relay services.
What I've been calling SMS is now "data sms" (or "nbs", in
code). There's a new way of inviting, called "User Sms" or somesuch,
that launches the user's SMS app with a URL and phone number, much like
email (save that no addressing is required by the inviter.) This new way
won't be involved in the ban on SMS permissions. (But play by NBS is
still awesome and will stay where it can.)
Get rid of the single sender thread and the complexity of querying
individual queues for their state (and the slowdown that happened when
successful writes had to wait for those to devices that weren't
responding). Instead each PacketAccumulator does its own waiting with
timeouts and backoffs, wakes itself when appropriate, and periodically
sends if it can make a connection. Now when there are a bunch of
messages ready they'll get sent pretty quickly once connections to the
remote device start to be successful.
A device where the OS isn't servicing enqueueWork() requests won't write
back to sockets it accept()s, causing the sender's writer to block
forever. So give it 30 seconds then kill the socket.
Add support for the old protocol, and define to use it. After the next
release has been adopted and everybody can read the new protocol another
release will turn it on.
Fix race condition that would have cleared all messages including those
added after a send began. In the process, move to storing outbound
messages individually rather than concatenated in their stream format.
Set a boolean every time app's upgraded, and clear it on first launch of
MainActivity. In between, if we try to launch the BTService but the OS
doesn't schedule it, post a notification asking user to launch the app.
Lots of changes. Old BT proto is no longer compatible. New one
batches messages per device and sends all on a single connection.
The queue (now a single buffer) is now a static global and the thread
that services it doesn't die (but gloms onto whatever Service instance
is most recent.) Code to pack and unpack the protocol can probably be
reused for wifi-direct or any other pass-through transport.
I'm seeing an assertion failure in cleanup that could be explained by
the cleanup happening when initGame() has been called but no subsequent
jni method that would have added the env to the map. So let's add it
immediately so that can't happen.
JNIThread is somehow sticking around sometimes and holding the lock for
a game so that that game can never be opened again. On the theory that
there's some place retain() was called but not release(), use the thing
in try-with-resources wherever possible. Which is pretty much
everywhere.
Also added age to the lock-holder report being uploaded.
I'm seeing something permanently lock a game so it can't be opened. So
add code to report the stack of the owner to Crashlytics when opening
fails 3 times in a row. It's stubbed out for non-debug builds.
I didn't like having to re-open a game to archive it after
rematching, so now there's a checkbox that lets you archive it after
either rematching OR just using OK to dismiss the alert. Also fixed
rematch of a solo game causing new game to be created in Archive
group if that's where the source game was.
squash me
Got tired of the space the forground-service notification icons were
taking. So now BT sends and receives are done via static threads and
onHandleWork(). The send thread times itself out quickly. The receive
thread doesn't yet. We'll see how long the OS lets it run and what
needs to be done to deal with that.
It's useful when testing to have the remote device play without human
interaction. So add an option, and UI to trigger it, for the players
created remotely in response to an invitation to be robots. There are
guards in place to catch the feature slipping into a release build.
Eliminate the writer thread, since onHandleWork() is already getting
called on a thread and can process the outbound queue itself. And get
rid of reader thread needing to wait for an instance of the RelayService
to be available to process a packet (which was taking 5-10 seconds all
the time): just post received packets via an enqueueWork() call using
the app context that's always available.
GameListItem was clearing its summary when kicking off a replacement
load. Instead keep the old around in case somebody wants it (e.g. to
figure out what menus to enable) until the reload finishes.
Also add logging of how long loading takes. I think a cache might be
called for.
Darned thing was dropping packets, failing to connect games built in
response to invitations, and otherwise misbehaving. First was due to not
resheduling when exited with outbound packets in queue; second to not
overriding relayNoConnProc() (due to signature change.) Though it still
happens occasionally.... Also added timestamps to track how long it
takes a packet to be sent and ACK'd.
It's too useful to live without. So: tap into the static list of live
Delegates, and if any of them has an Activity available use it to run
Toast on UI thread. Otherwise there will be no display.
My VSIZE is no longer legal, and apparently there's no workaround (no
way to safely figure the length of an array whose size is known at
compile time.) To avoid the risk of duplicating little constants, added
macros that define length in a way a new VSIZE can pick it up. Couldn't
make that work for struct field arrays, however, so there I'm using
constants.
Was assuming LIFO behavior, but with refcounting that's wrong and so I
was likely logging the wrong leaker in the case I'm trying to
duplicate. This simplifies the class while fixing that by tracking
owners as a Set.
Add a third processor type to the .so, and fix first compile-time
warnings and then a few dumb bugs based on assumptions about ptr sizes.
Works to play networked games and browse wordlists, but is not
extensively tested and needs to be before release.
Looking for a common but not-yet-reproducible situation where the
gamelock gets stuck preventing a game from being opened, usually related
to a move arriving. One thing that looked wrong is it seemed possible to
have the refcount drop to 0 then get increased again, causing somebody
to thing he has a lock when it's actually unlocked. So assert to try to
catch that case (and synchronize to make it much harder.)
command: find . -name '*.jp*g' -o -name '*.png' | xargs exiftool -all=
Did this because fdroid is complaining about metadata and I can't see
any obvious changes in the appearance of the app, but it's not
well-considered. Back this out if there are problems.
I think there's a rare race condition here. Assuming it results from a
bad network state rather than two sends coming too quickly, respond to
not having a connection by killing the thread so the next send will
retry.
Failure to immediately get a lock for a rowID in an intent meant that
other intent processing kicked in, and might have done the wrong
thing. So now if there's a rowid in it we try nothing else. And we use
the GameLock callback mechanism to deal with the case where it's
temporarily unavailable, as it will be when an incoming move is being
dispatched to it.
(DEBUG only) I'm seeing deadlocks in a non-reproducible way and need to
get at the source. This should log the stacks of long-held locks, and
notify via Toast (when possible) when it occurs.
No point in worrying about whether a thread exists without emptying its
queue. Instead have a thread start when something's added to the queue
and exit when it's been empty for a while. Only trick is the need to
remember the phone number[s] on behalf of which a retry's necessary
because jni code is waiting to see if combining can be done.
Should not show the new you-can-lookup-uncommitted-words hint for
already-played words, so needed to be able to tell difference between
the two. Now you can -- and on the gtk side I draw them differently
because I can.
It's crashing all the time and involves a system that only exists in
debug builds. So I should understand the root cause, but other stuff
needs to get done first.
Scrolling's needed in some cases, and you can't easily put a ListView
inside a ScrollView. So replace the ListView with a LinearLayout whose
contents I manage manually, and wrap the whole layout in a ScrollView.
Just keep a thread for each
phone we're sending to or receiving from, and pass them packets and
other things to send or (incoming case) process. So far the threads
don't die, but they will soon once there's nothing in their queues. Just
need to be sure there's no racing there.
This is to work around the frequent failure of the OS to pass enqeued work
into the service within a reasonable amount of time (I expect seconds, but see
delays of minutes or even hours.)
Theory is that the store's buggy site is presenting its
banned-permissions interface because it's detecting a Receiver that
requires SMS_RECEIVE permission even though the permissions themselves
aren't there any more.
I can't reproduce this, but while testing recovery from DatagramSocket
recreation in RelayService got enough to know that somehow a timer's
procptr's not set. So test before calling, and log instead of crashing.
Fix longstanding bug triggered by something as simple as putting device
into and out of airplane mode. Once the class-variable socket was
created it was never replaced. Now respond to exceptions that mean it's
useless and set it to null so existing logic will recreate it.
When enqueuing Intents for any of the three services, cache the
Intent. When onHandleWork() is called, remove the equivalent cached
intent. Then periodically check for intents that have been stored for
more than 60 seconds, and post a Notification alerting user to
stall. The user will likely have noticed that messages aren't flowing,
so this will simply explain the problem. Includes an "email author"
button.
Move some work into superclass where instrumentation can be added to
greater effect. And remove from BTService an earlier attempt (at stall
detection.)
toward tracking down relay stalls, log a bit more in resend process and
add UI to support an option where there's no reliance on
polling (i.e. FCM only)
I got bitten when creating a new variant leaving the permission out of
the manifest but using a config flag to claim it was there. Better to
have just one place to configure that. So check for presence in the
manifest and, for now, assert that that matches the isBanned flag. Soon
the isBanned flag can be removed.
Some people will decide to leave SMS comms in while I look for an
alternative. Maybe. Don't show them the same alert every time they open
the game. In the process removed what seemed to be a duplicate ivar.
There's no one way to launch an SMS app with a ready-to-send
invitation. So try the two that work on our two phone/app combinations,
iterating until startActivity() succeeds. The code's set up to have new
recipes added easily.
Lots of work dealing with banned permissions (SEND_SMS and
RECEIVE_SMS). First, if they're banned and NBSProxy is installed, just
use it as if the permissions had been granted. When it's not there,
explain at various points where users will otherwise be confused: when
they try to invite using data sms, or when they open a game that already
uses it.)
It's an error to use an env with the wrong thread, but it IS possible
for more than one thread to be associated with the same env. So don't
assert to the contrary.
Seeing the occasional crash and trying to plug leaks that might be
contributing. I think I fixed the Gameptr.finalize() assertions I've
been seeing forever but the thread match assertion remains. Logs will
help catch that.
Didn't think it was required, but once the app started getting messages
in the background it stoped working or crashed. Changes are minimal
because this service isn't responsible for receiving messages: it
doesn't have the long-lived threads of the BT and Relay services.
What I've been calling SMS is now "data sms" (or "nbs", in
code). There's a new way of inviting, called "User Sms" or somesuch,
that launches the user's SMS app with a URL and phone number, much like
email (save that no addressing is required by the inviter.) This new way
won't be involved in the ban on SMS permissions. (But play by NBS is
still awesome and will stay where it can.)
Get rid of the single sender thread and the complexity of querying
individual queues for their state (and the slowdown that happened when
successful writes had to wait for those to devices that weren't
responding). Instead each PacketAccumulator does its own waiting with
timeouts and backoffs, wakes itself when appropriate, and periodically
sends if it can make a connection. Now when there are a bunch of
messages ready they'll get sent pretty quickly once connections to the
remote device start to be successful.
A device where the OS isn't servicing enqueueWork() requests won't write
back to sockets it accept()s, causing the sender's writer to block
forever. So give it 30 seconds then kill the socket.
Add support for the old protocol, and define to use it. After the next
release has been adopted and everybody can read the new protocol another
release will turn it on.
Fix race condition that would have cleared all messages including those
added after a send began. In the process, move to storing outbound
messages individually rather than concatenated in their stream format.
Set a boolean every time app's upgraded, and clear it on first launch of
MainActivity. In between, if we try to launch the BTService but the OS
doesn't schedule it, post a notification asking user to launch the app.
Lots of changes. Old BT proto is no longer compatible. New one
batches messages per device and sends all on a single connection.
The queue (now a single buffer) is now a static global and the thread
that services it doesn't die (but gloms onto whatever Service instance
is most recent.) Code to pack and unpack the protocol can probably be
reused for wifi-direct or any other pass-through transport.
I'm seeing an assertion failure in cleanup that could be explained by
the cleanup happening when initGame() has been called but no subsequent
jni method that would have added the env to the map. So let's add it
immediately so that can't happen.
JNIThread is somehow sticking around sometimes and holding the lock for
a game so that that game can never be opened again. On the theory that
there's some place retain() was called but not release(), use the thing
in try-with-resources wherever possible. Which is pretty much
everywhere.
Also added age to the lock-holder report being uploaded.
I'm seeing something permanently lock a game so it can't be opened. So
add code to report the stack of the owner to Crashlytics when opening
fails 3 times in a row. It's stubbed out for non-debug builds.
I didn't like having to re-open a game to archive it after
rematching, so now there's a checkbox that lets you archive it after
either rematching OR just using OK to dismiss the alert. Also fixed
rematch of a solo game causing new game to be created in Archive
group if that's where the source game was.
squash me
Got tired of the space the forground-service notification icons were
taking. So now BT sends and receives are done via static threads and
onHandleWork(). The send thread times itself out quickly. The receive
thread doesn't yet. We'll see how long the OS lets it run and what
needs to be done to deal with that.
It's useful when testing to have the remote device play without human
interaction. So add an option, and UI to trigger it, for the players
created remotely in response to an invitation to be robots. There are
guards in place to catch the feature slipping into a release build.
Eliminate the writer thread, since onHandleWork() is already getting
called on a thread and can process the outbound queue itself. And get
rid of reader thread needing to wait for an instance of the RelayService
to be available to process a packet (which was taking 5-10 seconds all
the time): just post received packets via an enqueueWork() call using
the app context that's always available.
GameListItem was clearing its summary when kicking off a replacement
load. Instead keep the old around in case somebody wants it (e.g. to
figure out what menus to enable) until the reload finishes.
Also add logging of how long loading takes. I think a cache might be
called for.
Darned thing was dropping packets, failing to connect games built in
response to invitations, and otherwise misbehaving. First was due to not
resheduling when exited with outbound packets in queue; second to not
overriding relayNoConnProc() (due to signature change.) Though it still
happens occasionally.... Also added timestamps to track how long it
takes a packet to be sent and ACK'd.
It's too useful to live without. So: tap into the static list of live
Delegates, and if any of them has an Activity available use it to run
Toast on UI thread. Otherwise there will be no display.
My VSIZE is no longer legal, and apparently there's no workaround (no
way to safely figure the length of an array whose size is known at
compile time.) To avoid the risk of duplicating little constants, added
macros that define length in a way a new VSIZE can pick it up. Couldn't
make that work for struct field arrays, however, so there I'm using
constants.
Looks like certain launches that went through DispatchNotify could
create a second "instance" of the app. Giving that Activity a launchMode
of singleTask fixes this (singleTop does not.)
Launching a game in response to a wordlist it needing being downloaded
went through a different Intent-creation path from the rest, and
resulted in a second instance of the app showing in the launcher. This
shoudl fix that. I don't *think* it's the only way I'm getting second
instances or stacked GamesLists, but it should get fixed.
I modified existing translated strings, adding the new clause. Not doing
that would cause the existing strings to be stripped because they'd no
longer have the matching set of format specifiers.
Fix a race condition introduced by making initing jnithread be
asynchronous: all the layout that was supposed to happen after listeners
were added instead happened before, so that they weren't there to be
installed as part of layout. So now, after adding them, get them hooked
up to the UI. (The complexity of this is all historical: at some point
the listeners were getting added BEFORE there were views to attach them
to, so they were cached and added later. Probably now they could simply
be installed as part of adding them. But I'm not doing that now.)
Was seeing frequent failure to open games as JNIThread.getRetained() was
unable to get a lock without waiting. Which it can't do on UI thread. So
added a method to GameLock that takes a callback to be called when the
lock's obtained, with the actual waiting done on a local thread. Then
fixed BoardDelegate a bit to not crash while waiting for the callback.
Once built for SDK 26 the old GCM support code crashes as it's calling
startService() from background. Duh! Will have to bring in Firebase's
replacement at some point. For now if I'm to release it has to be
without GCM.
more dev-only code, likely: if the game that's sending the invitation is
open when the invitation arrives, should still check if it's a dupe or
should be accepted.
Was creating a second game in some cases in response to duplicated
invitations. Add a test for forceChannel: if there's already a game with
the same gameID *and* forceChannel assume the invitation is a dupe and
drop it.
So I don't have to open a game to see e.g. how it's connecting, refactor
the network status stuff so it can be called from both board and
gameslist views. Then add new context menuitem that calls it.
Instead, keep them forever (for now), sorted by how long since they were
last seen. A Delete button's probably needed to prevent ex-partners from
sticking around too long. :-)
Adding a second permission a while back resulted in two Actions being
received and acted on. Needed instead to group the two and receive a
single Action IFF both were granted.
Put up an error message if too many tiles selected for trade
(a condition that couldn't exist when the pool was guaranteed to
have at least 7 in it.) (It's a hack: there's not even an enum
giving Spanish's code, and the lang_locale stuff in info.txt isn't
making it into the .xwd format.)
I've learned a bit about java since writing that class. Should be faster
now without the sleep/polling. Also making it illegal to use blocking
lock() call on UI thread. There may be some assertions to fix in the
next few days.
Was seeing race condition on slowest device where JNIThread init hadn't
finished before the event loop tried to use the gamePtr. So use
wait()/notifyAll() to fix.
And: Don't start the foreground service with its user-visible
notification, as that's too obtrusive and the ACL_CONNECTED stuff seems
to work well enough. Launching timers on a new install is mostly for my
own dev use, but won't hurt user experience either.
On O and beyond it's possible to tweak notification channels separately,
meaning users can hide the new BT-is-running notification and still get
game event notifications. So add a button that takes you to the right
Settings app page.
Separate processing of sockets from accepting them so that when an ACL
CONN notification is received and we open a socket (but don't yet have a
Service running because the ACL thing is most likely for some other app)
we can set it aside to be processed once we do have a service. Use the
same block-until-non-null thing as in RelayService to keep that thread
free of NPEs.
Rather than just dropping it and going back to a wait that will likely
last forever. I *think* wait() throwing that exception means the
thread's being killed, in which case the exception should be thrown up
to the containing loop that will then exit.
I thought I had to stop using a service instance before returning from
its onDestroy(), but that made the UI incredibly laggy AND appears not
to be necessary. At least in a bit of testing things still work.
Oreo's creating a new service instance for every single intent passed to
enqueueWork, meaning a brand new set of threads with a new queue, empty
set of messages to be retried, etc. was created every time, and all
attempts to optimize and retry were broken. So: make the threads a
static singleton that are given a Service instance from onCreate() and
told to drop it from onDestroy(). The threads proceed until they need an
instance, then block until one's available. Seems to work on Oreo and an
older Android as well.
Apparently one of the newer Android SDK levels adds the requirement to
have RECEIVE_SMS permission in order for a broadcast receiver to get
called. Meaning receipt didn't work even if SEND_SMS had been
granted. Since they're both in the same group (for now) the OS will
grant the second silently if the first has been granted, but it still
has to be requested. So request both at the same time. This still leaves
the problem that a user who's never tried to create an SMS game won't
have been asked for either permission and so won't receive SMS
invitations, but fixing that is for a later release.
Other code will take care of duplicates. This was meant to avoid a race
condition, for which 5 seconds is enough. Blocking forever complicates
testing.
Receiver is created and installed by a non-UI thread sometimes so can't
create the Handler there. onReceive() is called on an ok thread however
so create it there on demand. Fixes crash that showed when receiving
relay messages in background.
Fire up the receiver thread, and start the service, on receipt of this
ACTION (if they're not already running.) On start, the service takes
over the thread and begins dispatching messages. Works to launch the app
when it's not running and in most cases, though messages received before
the service launches are currently dropped, things seem to work.
reset timer on receiving meaningful data and on moving app to
background. If it fires and we're in background, kill the service.
(There's currently no way to restart it except bringing the app into the
foreground. Fixing that's coming.)
Moving toward a better BT invite experience: use BTService to scan for
ourselves on all paired devices, and only allow selecting from among
those on which we're running (and so likely to respond to an
invitation.)
It's a pain to have to change code to run in Genymotion, and to have
upgrade not testable using Genymotion. Consider verting this change
before next release.
Did a bunch of stuff to inherit from JobIntentService and use enqueue(),
but doesn't work yet. OS is unable to bind, with this error:
09-21 17:20:51.678 3050 3050 W JobServiceContext: Time-out while trying to bind 2edee28 #u0a277/1111 org.eehouse.android.xw4dbg/org.eehouse.android.xw4.BTService, dropping.
Notifications don't work on Oreo without this change, which includes a
new Support Library in order to get NotificationChannel and creates and
uses that as docs describe. Requires that MinSDK be raised from 8 to 14,
which may lock some users out. It *should* be possible not to do this in
the fdroid variant since their app store doesn't requires SDK 26, but
I'll look at that later.
For some reason my laptop wouldn't build without this change. No idea
what happened to the newer version I was using or if the change
works (beyond compiling). Should be easy to find the change later if
it's a problem.
Some devices, including my Moto, are apparently calling server_do() more
than most. When the game's supposed to be asking the user to pick tiles
that resulted in stacked TilePickAlerts. The stack of these
sending (taken together) too many picked tiles to the game made it
crash. So modify server to have only one pending tile-pick request going
at a time. Because the server can't know when the user dismisses the
alert in Android and so won't post again, respond to the dismissal by
closing the game. Reopening will put it in a state where the tile picker
can get called again.
Logic error meant it was never sent. Now always send on receipt of a
well-formed invitation, even if e.g. the recipient's missing a wordlist
and the game can't be started immediately.
Had inadventently changed how NetLaunchInfo was transmitted, and crashed
on receiving from older builds. Fix to not crash, and then fix to send
and recieve in the old format.
So now all jni code uses a single dutil context, but also a single
mempool and jniutil instance instead of new instances of the latter two
per game and dict-iteration.
A number of jni calls were "stateless", which meant they allocated their
own vtmgr and mpool instances each time invoked. Instead invoke them
with the global jni closure and add to it vtmgr (already has mpool) and
use these instead of allocating/freeing each time. To make sure no race
conditions are introduced (mpool, though debug-only, is probably not
thread-safe), guard these new uses with an in-use flag. If that fires
I'll need a mutex or something.
Somehow I got a wordlist into a location different from what was
recorded in the DB table and since the delete command matched on
location as well as name it was never deleted (which meant the checksum
was never updated and so upgrading never seemed to succeed.) Removing
the match on location fixed that problem, and since I don't see any harm
in cacheing only one version of a wordlist will simply leave it that
way.
Did a bunch of cleanup as well.
I'm seeing several IllegalStateException crashes due to e.g. having an
alert posted when app's in the background. Need to fix them, but the
debug build crashing isn't helpful. Log.e() instead.
The pesky thing is back. When app's in the background with an
unconnected game open displaying the "resend/wait" alert and the game
connects get IllegalStateException because the fragment stack's being
modified after onSaveInstanceState() (or, because the dialog fragment,
saved as an instance variable in BoardDelegate, dates from an earlier
state. Anyway, catching and dropping the exception and elsewhere failing
silently to rebuild the alert seems to fix the problem, but the right
fix is likely different. I suspect hanging onto that iVar is wrong, and
that the dialog should go away when onStop() is called and then be
rebuilt later from saved state. But for now, not crashing is good.
Remove a bunch of duplicated code, replacing with implementation in
XWService. Fixes duplicate invitation and game opening policies being
slightly different.
Try some funky layout shite to get, within a horizontal linear layout,
the first text field trucated if necessary so that the second (holding
the score) can be fully displayed. Tested on exactly one emulator so
far.
Try some funky layout shite to get, within a horizontal linear layout,
the first text field trucated if necessary so that the second (holding
the score) can be fully displayed. Tested on exactly one emulator so
far.
I think there's a bug in Weblate because I've seen this before: where
the English provides only an <other> the translation comes back with
only a <one>. That's wrong. Try adding a <one> in the English case to
see if that makes a difference.
I think there's a bug in Weblate because I've seen this before: where
the English provides only an <other> the translation comes back with
only a <one>. That's wrong. Try adding a <one> in the English case to
see if that makes a difference.
Otherwise it takes too long to scroll if you have hundreds of
games. To make this work had to move scroller to left side of games
list display as otherwise the scroller steals events from the expander
thingies.
Likely because of something in the jni world unset per-player dict name
is empty string rather than null, so test for that too. Fixes dicts
popup in newly-created game have an empty first line " (in use)".
I like this better than the previous fix: rather than share a
thread->env map with the game world allocate a new one for each
iterator. This could cause problem if the iterator is used on threads
that don't currently call map_thread(), or if there are callbacks that
need to look up the env that I'm not aware of. Needs testing...
Stumbled on a NPE opening up the wordlist browser configuring the first
game on a new install. So now test for null there and init early if
necessary. Seems to work, and won't do anything in places were not
needed.
Remake the min and max spinners every time either value changes so they
can't be used to set nonsensical values. (Which leads to immediate
crashes.) I'm sure this wasn't always a problem, but...
As reported to google, dict iterator destruction was crashing due to a
race condition if it happened after a game using the same dict had been
closed since it needed a mapped env that the game closure would
remove. Fixed in two ways, one by adding the mapping prior to the code
that uses it (a common pattern: add happens many times, whenver it might
be needed, but remove only once), and second by passing env into the
code that was crashing.
The mapping stuff remains inherently racy and I'm not sure now how to
fix that. It depends on there being a place to unmap after which it's
guaranteed the mapped value won't be needed again. When two
objects (game and dict_iter in this case) map the same env/thread combo
there's a race.
Looks like the assertion was left in when adding support for dual-pane
mode, as all other onPosButton() implementations called super rather
than assert. Which this one does now too.
Getting ANRs because (I think) the main thread's waiting for the write
thread to die and now the write thread's doing a ton of work
sometimes. So move the threads into a standalone object that can be
allowed to die on its own time without anybody waiting.
I *think* the reason I'm occasionally seeing toasts about not finding a
move is that when the engine's interrupted by there being a UI event in
the queue that error is posted. Instead try posting only when at the end
of the search nothing's been found.
Having reconfigured to use non-existent relay port as a test of falling
back to the web apis, tweak stuff: send the packets that have been
accumulated when an EOQ is found (rather than dropping all of them
immediately) before exiting the write thread; and start the threads up
when posting a packet in case they aren't (they may not be when the post
happens via timer firing.)
Seemed to be causing ANRs. Integrate instead into outgoing message queue
by using poll(timeout) then checking for unack'd packets every time
through the loop (but not more than once/3 seconds or so.)
Presence of timestamp instead of a boolean determines whether packet
should next via Web. Timestamps might also allow to process a larger
number of unacked packets in a single timer fire....
Track ack'd and unack'd packets. When there are ten more of the latter,
skip the UDP-send step. This is probably not the algorithm I'll settle
on (an explicit PING to the relay over UDP might be simpler), but it's
simple and easy.
Send each packet via UDP if that's thought to be working (always is,
now) and start a 10-second timer. If it hasn't been ack'd by then,
resend via Web API. Tested by configuring to use a UDP socket that the
relay isn't listening on. Only problem is that the backoff timers are
broken: never stops sending every few seconds.
The plan's to use the native relay protocol first, then to fall back to
the slower but more reliable (esp. on paranoid/block-everything wifi
networks) webAPI. This is the name change without behavior
change (except that the native kill() to report deleted games is gone.)
When grouping to allow multiple packets per outbound API call I forgot
that some are there to mark the end-of-queue: can't be sent! Trying
caused a NPE. Now if any EOQ is found in the queue that batch is dropped
and the thread's exited.
Making the right_side elem match its parent height prevents the
lower-right region of game list items from falling through and
triggering a toggle-selection event.
Making the right_side elem match its parent height prevents the
lower-right region of game list items from falling through and
triggering a toggle-selection event.
As expected, moves are no longer received instantly because the UDP
socket isn't available for the relay to write too once the URL
handler (relay.py) finishes.
Ideally the comms module wouldn't go through its connecting routine in
order to join a game. To that end I added a join() method to relay.py
and code to call it. Joins happen (pairing games, starting new ones,
etc.), but after that communication doesn't. First part of fixing that
would be to make cookieID persistent and transmit it back with the rest
of what join sends (since it's used by all the messages currently sent
in a connected state), but I suspect there's more to be done, and even
that requires a fair number of changes on the relay side. So all that's
wrapped in #ifdef RELAY_VIA_HTTP (and turned off.)
Was coming in as a string when called via curl. This may be a problem
for other ints if I go with lots of params instead of a json, which is
looking less likely.
join's how devices will create or join existing games. It more
compilicated than I'd like but seems to work except that once a slot's
assigned it's unavailable to anybody else even if the other fails ever
to respond (i.e. needs the ACK function of the c++ relay.)
ACK doesn't need to wait 2 seconds for a reply, and when it does so the
next send waits too. Eventually we'll want to combine messages already
in the queue into a single send. For now, this makes things better.
So far uses curl and json-c to send b64-encoded data to new script
which is able to echo the data. Next that script will need to open a
UDP socket to the relay and return results that appear before timeout.
Working around there being a border around the game-type image area.
With this change long-tapping works only on the right 2/3 of the
region. There might be a fix, but it's still better than there being
a hole (the border) in the thing where behavior's different.
Duh. The .java file was removed, but the declaration that all apps can
handle an intent (that requires WakeLock they no longer all have) was
not. Should fix crashes I'm seeing.
Board and Games List were using same menuid which meant that even when
chosen from Board's menu it would up getting handled by GamesList (in
dual-pane mode.)
Somehow the d variant was crashing without the WAKELOCK permission, the
OS having invokes something GCM-related. This should ensure nothing
GCM-related can ever happen.
This seems to fix that app, when built by me where GCM_SENDER_ID's set
in the environment, being a battery hog. Apparently google's code
doesn't handle being passed the wrong senderID very well.
I'd added inputType="text" everywhere I added maxLines="1" but it turns
out that breaks touches being handled at least on some devices. And it
makes no sense to have an inputType for something user can't put into.
Several EditText fields are configured so monkey breaks things, e.g. by
entering too much text in the default player name config. Fix. This
probably won't impact users but it lets the monkey tests move on to
other things that might.
Can't repro on Nexus emulator running 4.4 nor on Samsung running 4.4.4,
but the reporter says this fixes it. And from reading it appears
expecting older devices to load Material themes without an AppCompat
library is wrong.
For the ORDER BY clause that governs how games are displayed within a
group, use a static string built from a list of clauses that are then
easy to move up and down. Add clause that moves games with unread chat
to the top. Another commit will modify the display so it's clear why
it's there.
Bring back the test for being null, and add an additional one via a new
boolean iVar that we haven't tried posting it already. On some devices
there's enough of a lag that the is-null test passes several times
before the first makeDialog() call, leading to a cascade of attempts to
create that hangs the UI. The test's needed, but only once should be
start the process.
When the back stack is restored commit() is not called, so the fragment
needs to save it. Without this 0 is passed to popBackStack() and
everything's dismissed, not just the one fragment.
Rather than just pop whatever's on top of the back stack, which might be
the alert whose listener called finish(), use saved commit() ids to pop
down to the fragment itself. This feels like a risky change, and it's
tested only by back-button-dismissing the Wait alert in an unconnected
game in BoardDelegate, so needs some bake time.
When SDK >= 19 there's an API to tell if a listview is full enough to
require scrolling. So use that, rather than the raw count of games, to
decide whether to offer new users to hide the new game buttons.
What's here is merged with the main one, so no point in duplicating and
it's just a pain to maintain. Remove as much as possible that's not
unique to the debug variant (crashlytics and wifidirect, basically.)
Oops. Debug build assertion showed I was leaving an infrequently used
field out of serialization, adn then required that that object be
Serializable and implement equals() to pass more tests.
Get rid of treating 0 as a legit date (1970 being illegit). Tweak
formatting. It's not perfect, but few people see it so we'll see how it
goes during development.
revert a bit, dropping use of git revision and repo to provide an
order. Instead use aapt (which is an ubuntu package now) to pull the
version code and appID from .apk files, order by versionCode, and
secondarily by file mod time.
Auto-update was based on my manually setting what the newest is. Better
to use the git revision stored in the .apk, or failing that (later) in
the file name, to determine "age". This is all based on forcing a linear
order on git commits, but at least at the granularity of releases that
should be ok.
Mistake was putting "download more languages" at the bottom of list of
wordlists in one language. Having separate strings is a bigger change,
and in context just "Download more" works in both cases. Change name of
string so translators will notice.
Was passing through DlgState params a number of classes that weren't
serializable or that didn't have equals() methods required for DEBUG
assertions to pass. Added versions of equals() that just call super in
non-DEBUG case since it's a lot of code and isn't required except for
testing. (Serialization of course is.)
Now that params[] are being bundled all objects passed that way must be
serializeable. And as long as I'm asserting bundle success using
equals() the objects being serialized must implement it.
I wasn't saving/restoring params[] Object array with DlgState, and under
circumstances I can't reproduce that could cause a null object
dereference I was via the Play store. So added a test case that failed
for DlgState instances that have non-null params arrays, and fixed that.
Attempt to fix invite alert sticking around after game connection
finishes, which resulted from the variable pointing to the alert having
been set to null. So stop setting it in onDismiss listener. I think
trying to track it via an iVar is a mistake, that dismiss by tag or
somesuch is the right way to go. But my dialog fragment code isn't up to
that yet.
Add new Utils methods that turn Serializable objects into B64-encoded
Strings, and vice versa. Use in place of existing code, and use anew to
store the array of DeviceID records. Implement the "Delete checked" button.
In Dbg version only (with enabling boolean moved to BuildConfig.)
Currently crashes when first used, and recipient can't always connect to
relay until app rebooted. And of course there are still open questions
like how to populate the list -- how user learns the deviceID of an
opponent.
Was previously a "global" so that a resend over SMS might delay a resend
for relay games. Since communication problems tend to be
comms-type-specific that's dumb.
Added a wrapper function since for compatibility with the jni's encoding
the flags matter and need to be the same everywhere. Or at least there
should be no chance of their getting changed.
As a first step, use mine and android's and assert results are the
same. Note using built-in Base64 class requires moving the min android
version from 7 to 8.
Copy config file into the right place, turn it on for debug builds, and
fix compile-time complaints by adding -keep clauses. Resulting builds
seem to work (after a few minutes in emulator only) except that net
status icon taps crash at first (eventually recovering somehow) likely
because of problems with base64 en/decoding which goes through jni.
Get rid of black background of center icon, instead running the green or
red all the way to the middle. For better contrast make the green and
red darker and the non-active arrows white.
The debug-only settings for disabling individual conn types weren't
kept in sync with which types were enabled, and that set of types was
lost on rotation. Fixed (including keeping the disableds across rotations.)
Since it's now possible to add a name for an SMS contact added manually,
don't assuming a missing name means it was added manually. Just leave
blank if not provided.
I made these changes a while ago then stashed them. They worked when I
needed them then, and seem to work now, but I haven't verified now so
don't trust 'em too far. :-)
Crittercism is dropping free support of FOSS apps, so I'm dropping
it. Unfortunatly integration isn't as clean: so far at least I haven't
figured out how to make it possible for others to build the 'd' variant,
which includes Crashlytics, without their having a Fabric API key.
Resetting it on every ACK etc meant it happened forever at the most
frequent interval, at least on devices that have both sides of a game,
which is my usual test setup.
In game config, when the play game button's hit and the changes pending don't
"matter" (require reset), apply them. The motiviation is to allow
setting the "disableds" and immediately launching a game via the button,
but I think it's the right thing to do anyway. Needs testing.
Deal with occasional droppage of SMS data messages by running a
timer (via AlarmManager) with backoff that resends any pending moves for
games connected via SMS. Successful receipt of SMS data resets the backoff.
cleanup, iterating over built-in data rather than my hard-coded VARIANTS
and BUILDS arrays, which now go away. Renamed obj- and libs- directories
to better fit the variables the system provides.
Add DEBUG-builds-only UI at the bottom of GameConfig for turning on the
comms feature dropping outgoing or incoming messages. The idea's to use
this to reproduce and fix stalls. There's a noticable slowdown opening
the GameConfig fragment, but it should all be no-op in a release build.
Add BuildConfig boolean that's true when a gradle property is passed in
from the fdroid metadata/build file. Test when putting together upgrade
state and omit app info when it's true, same as if from Play store.
F-droid build system modifies git-controlled files so no build will ever
be clean. So look for a -P property and if that's set don't use the
--dirty flag when capturing the git revision. Requires a change to the
fdroid metadata to pass that property.
Years back what I read said you opened and closed the database around
every query. Now I'm reading differently, that the OS will close it for
you on app shutdown and that it's ok to just leave it open. Trying
that after a few minutes on one device looks ok. Will need to test the
heck out of it, especially on older OS versions, before ship.
Play store reports crashes but stack crawl doesn't tell me where
from. So try catching the exception and on debug builds logging what
dialog is responsible.
When those became an advanced feature I added a warning for upgrading
players who'd miss them. That was long ago, and the warning was only
supposed to last a release or two.
showDialog() goes away except for PrefsActivity which can't do
fragments. Move stuff required by PrefsDelegate into it from
DelegateBase since no other subclasses does old-style Dialogs
any more. Remove a bunch of stuff from DlgDelegate, e.g. saving
state, that no longer gets used.
Was getting crash with "Can not perform this action after
onSaveInstanceState". This lets the back stack get back into shape so it
can put up another alert.
Add to DldID the ability to specify whether dialog fragments bearing
that id go on the back stack. Set INVITE alerts (only) to use that
mechanism. Having them on the back stack seems to be required by their
buttons being hacked to not dismiss them, which in turn is required
because I want them to stick around underneath other alerts their
buttons launch.
Get rid of explicit enable-dualpane boolean pref, instead relying on the
three way default/force-phone/force-tablet pref. Offer to change that
rather than the former in BoardDelegate when run on upgrade. Move the
prefs setting from Debug to Appearance since it's "real" now.
Recent recommendations are not to attempt to handle it myself, so don't
and we'll see how it goes. And unify notions of tablet and enabling
dual-pane: a device that isn't dual-pane should not be using the new
single-activity stuff at all.
Had to disable use of the back-stack for DialogFragments, though that
means I can't prevent duplicates from stacking up (esp. in the
pathological robot-vs-robot case, but also just when I rotate the device
while a "rematch" alert is up.) The problem seems to be in dismiss
actions being handled too late. One easy-to-duplicate case is the
tile-picker. When it's enabled and you commit a turn I post first the
confirmation and then, in response to a "yes", the picker. But the
picker gets added to the stack moments before the confirmation is
removed, and it's the nature of stack removal that everything above
what's being removed gets pulled too. So you never see the
picker. Simply post()ing a runnable to put the picker up later fixes
this one case, but a similar trick doesn't work elsewhere, so I'm
punting until I have time to root-cause the problem.
Each alert gets a unique name. Add to backstack, and before doing so
look for another with the same name and dismiss it. This works most of
the time, especially to prevent them from piling up with orientation
changes. But in a robot-vs-robot game, and occasionally in a game with
three robots and one not, SCORE alerts stack up. It's the removal that's
failing, not the test for a duplicate, so perhaps I need to somehow wait
for it to succeed before adding the new alert.
DELETE_DICT Action is expected to flow through child, at least until I
fix to disable deleting a dict while it's open in the dict browser. It
may be confusing but with the assert gone does no harm.
Make the default be that it's on, and move the new-feature notice from
app launch to board open, the point where most will see the feature for
the first time. Remove the notify-and-exit thing, which is harder to
make work from the board, instead just posting an alert, if user chooses
to disable, that the setting change takes effect after a restart.
Needs to show different text and buttons, so now that it isn't getting
dismissed and rebuilt every time a button's pressed force that process
manually the time it's required.
When game's resumed (brought from background) and Wait alert is up
onStart() is called before getDialog() has a non-null value to
return. onResume() is the place to modify the dialog.
Rewrote Wait/Invite alert to use new functionality that allows it to set
its AlertDialog buttons to not exit when tapped. Turns out things are
much simpler when I don't have to hook into all the places my dialog
might need to be put up again.
Override onCancel() in order to finish when the alert's closed via the
back button. Count the number of alerts and when it drops to 0 (when all
the alerts launched from the invite/wait dialog are gone) put the
invite/wait up again.
I can't figure out how to prevent rotation from duplicating the wait
alert, and don't have the ability to have makeDialog() return null, so
when that happens -- when a second is being created -- dismiss the first
first. And clean up some callbacks made unnecessary by onDismiss().
Try to track in local saved state enough, including a count of
sub-alerts the main Invite alert has put up, whether when something's
dismissed the alert should go back up or the game should be
closed (which is the right thing to do when the back button's pressed.)
Hack: setting a variable to null in onSaveInstanceState() then testing
it in an onDismiss() call that happens as part of rotation avoids an
exception when rotation the phone.
Move most state that needs to be saved in Bundle into a single
Serializable class so adding new is easier. Add to that the current
selected group so we don't crash when nothing's been set. (I think in
the move to DialogFragments I lost the code that would have disabled the
Move button. That's back.)
Move call to end of derived class' method. This may very well be the
solution to a bunch of random crashes logging
"java.lang.IllegalStateException: Can not perform this action after
onSaveInstanceState". Chief suspect is the change in XWActivity.java,
but I'm making it everywhere for consistency's sake: it's a good habit
to get into.
New classes implement custom alert and its view, where most of the logic
for putting up one button per tile, hiding and showing buttons based on
what's left, etc. lives. Rough, but works well until rotated, when gets
redrawn without spaces for the buttons that could come back.
The crazy hack to get around Android's lack of modal dialogs is at long
last gone. Remove semaphore and code that uses it. Will leave the
background thread in place, at least for now, as removing it would be a
huge change.
Make face-up tile picker util method return void and add mechanism for
passing in selected tiles asynchronously, as has been done recently with
the rest of the once-blocking util callbacks. Works perfectly in the gtk
case. Likely crashes in curses (if picking face-up option is on.) In
java all the callbacks are there but rather than put up a UI we pretend
the user says "pick 'em for me" each time. Putting up a UI is next.
Rewrite the overly-complex invite dialog that's posted when a game opens
and is missing players so that it doesn't crash after an orientation
change. It's still possible to dismiss it yet have the game stay open.
The goal is that a game in that state always has a dialog on top of it
so you don't get confused about e.g. why you don't have tiles, but
there's more work to be done there.
Also added a common superclass for all my DialogFragments. It may be
useful e.g. for having each DelegateBase instance able to track what
dialogs it currently has up, e.g. to have a policy about how many of the
same class can be live. I'm thinking here of how to prevent a
robot-vs-robot game from ending with a stack of 25 move reports. (The
change here is that the old activity.showDialog(int) only allowed one at
a time, and some of my logic takes that for granted.)
There were multiple places launching it, which in a DialogFragment
world (in the absence of other per-dlg-id controls) means stacked
multiple alerts. And dismissDialog() didn't work. So use a boolean to
prevent more than one being posted and keep a reference to the
DialogFragment behind it in order to be able to call dismiss() on
it. That's the only place I was calling dismissDialog() (other
than the tile-picking stuff which has to be rewritten anyway) so this
little hack is probably ok.
This alert still gets messed up with the screen config changes. Fixing
that is next.
I can't get preference fragments working in my world (mostly because
nested PreferenceScreens don't open when clicked on and the workarounds
for that are too complex) so next best thing is to fix alerts to work on
top of prefs activity. That involves creating a new transparent activity
that subclasses XWActivity and has a DelegateBase subclass. Launching an
alert is then just a matter of passing DlgState into the new activity
via an Intent, at which point everything Just Works. Button clicks are
returned via an intent as well.
Continue conversion of alerts that required blocking the JNI thread. Now
board_commitTurn() takes a second boolean indicating whether phonies
found have been approved by user. Common code informs user, and if he
approves client code calls board_commitTurn() again. In case where
turn's lost there's no call to make back, but there's the undesirable
change that if a robot moves next its move will be reported on top of
the turn-lost alert. Ideally new alerts would appear under, not on top
of, those that have not yet been dismissed.
Next step in converting util_ methods that required blocking: blank tile
assignment. Now post a query and add a method that the client code can
call when the user's decided. Include enough state (col, row, and
playerNum) so that it's probably pretty safe.
Probably breaks curses build, but for gtk and Android
turn move and trade confirmation into a two-step process, making
board_commitTurn() non-interactive when called with a second
parameter. The old blocking util methods now return void and it's up to
the client code to interact (on the main thread) then re-call
board_commitTurn() if appropriate.
First attempt to stop blocking the jni thread: instead of returning a
password from a util_ method, have it include enough state that the UI
can return, put up a dialog, and then pass that state and the password
back and have them matched up again. I think this will work for the
remaining blocking Alerts too.
Includes a hack in DBAlert that's required I think because
GameConfigDelegate is launched "for result". onCreateDialog() always
fails after a rotation the first time because GameConfigDelegate isn't
there to have its makeDialog() dispatched to. So it puts up a dummy
alert and then post()s code that is successful in calling makeDialog()
to get an alert from GameConfigDelegate that can replace the
dummy. Nothing shows on the screen on simulator anyway.
The major problem remaining is that blocking alerts in BoardDelegate are
recreated after rotation but the thread that was blocking has been freed
so nothing can be done after the new alert returns. E.g. blank tile
picker will be posted again, user will pick a tile, but the common
code's not in a state to do anything with that choice (which cannot even
be "returned.") Options are to find a way to make the JNIThread survive
the configuration change without unblocking or to rewrite all the common
code to not expect return values from util_ methods.
This commit is not well tested, and diffs don't allow a thorough check
of the conversion of each DlgID type.
New way of adding onDismiss listener to DialogFragment-based alerts
required API change to DBAlert and the makeDialog() method. Fixes
customizing default player name.
Had left out dispatch of "negative" buttons populated from an
ActionPair. Tested for offer to remove redundant new game buttons at
bottom of games list.
For GamesListDelegate only so far, replace calling <activity>.show(int)
to launch an Alert with something producing and showing a
DialogFragment. Replaces passing and saving state inside the
DelegateBase subclass with saving it as part of the fragment's bundle,
and it looks as if a single class will work for nearly all of the alerts
managed by DelegateBase.{onCreateDialog,prepareDialog}(), which will
eventually go away. The beauty is that the implementations of
onCreateDialog and onPrepareDialog remain, but as the body of a new
makeDialog() that's called by the fragment's onCreateView. Less code
changes, but now it's all called every time an alert's created.
Since associated DlgDelegate instance doesn't survive a rotation, needed
to pipe results through a DlgClickNotify implementation on XWActivity
that then dispatches in the DualpaneDelegate case to all visible
fragments. I hope this turns out to work for all DlgClickNotify
implementations as I switch them over. We'll see.
Add script that when copied to device and run from 'adb shell' captures
all app and OS logs to file. Next: add code to XWApp.java to copy it
from assets/ and run it on launch.
On 2.3 anyway there's a java.lang.VerifyError trying to load that
class. So wrap the loading in a new class that catches the error and
sets a boolean so subsequent calls don't need it. There are other calls
that could load the class, but they may not be reachable. To be tested.
Required to fix alerts going away now that orientations are handled by
the OS (which in turn is the preferred way to do things now since not
ALL config changes CAN be handled by the OS.) With this change some
alerts will probably cause crashes during a rotation. Those all need to
be fixed.
Add listener interface to DBUtils and hook StudyListDelegate into it so
that if a word is added while the list is being displayed the new word
shows up immediately.
wordlist browser selection and downloaded info about downloadable
wordlists didn't survive rotation. They do now, the latter as a huge
serialized array. To make selection work I save the keySet() of a
mapping of selected names to the views that represent them. Now the
presence of a key, even if the value is (temporarily) null, signals that
something's selected.
Old code that didn't rotate properly was meant to associate view IDs
with fragments so I could iterate over them, e.g. to dispatch
intents. To replace that the common superclass of all fragments now
keeps a set of active ones and provides a method that uses that to find
the fragment that owns a view. So I can iterate over fragments based on
the dualcontainer-contained views as before.
Remove the generated FrameLayout that was breaking restoration after a
config change. Set layout_weight=1 for all fragment root views so they
get half the screen in landscape mode. Remove some code. Problems
remain, first among them that notification intents aren't dispatched correctly.
A group that's closed stays closed even if a game's added to it. The
change flushed out a bug where the groups cache wasn't being invalided
on a game move, so fix that too.
Name file and set internal constant sent to server to use same git rev
string so that the script will correctly identify whether the version I
have is the latest it has available. Fixes server offering to replace
with what I already have.
replace symlink silliness ant required with gradle commands to sign
debug builds with the checked-in (local file) debug keystore. Make
it possible for an environment variable to override in case somebody
wants to use his own.
Looks like there are battery drainage issues with the current half-done
implementation, so add option to turn it off. Only takes effect on
restart, and only matters on dbg variant now since it's compile-time
disabled on the main variant.
Also add discovery machine state to ConnState printout, stop running the
machine over and over (likely the battery problem), but add ability to
restart it triggered by the ConnState's "reconnect" button.
Also start persisting the most recently seen set of peers. If service
discovery and not connecting is the problem having these available to
try to connect to on startup might help.
Build different jni code (into different obj and libs directories) for
the different release/debug and xw4/xw4dbg combinations. This works when
only one build is done, but when doing two some later task tries to put
both multiple same-named libraries into the .apk and so fails. I still
think it's worth adding this to keep from using the wrong binary, but
that needs to be fixed.
cleanNDK just nukes all the directories. Easier than invoking build-ndk
clean...
Merging of AndroidManifest.xml files meant the dbg flavor was getting
its C2D_MESSAGE permission and the main flavor's, which on recent OS
versions meant it couldn't be installed. Use substitution from gradle to
fix.
GamesList menu was in a bad state after a game closed because hadn't
been rebuilt. It's simplest to invalidate it surgically, when the
BoardDelegate notifies that the board's closed, though doing in in
onWindowFocusChanged() might catch more edge cases. We'll see. I want a
low-risk impact right now.
but not on emulator running same OS version (or close). And in spite of
an exception being thrown the work being done, hiding a preference,
still succeeds. Whatever.
Fixes case where user has received invitations but not allowed SMS then
decides to do so. It's a hack giving Perms23 class knowledge of how to
send moves etc. Instead I should be letting interested parties register
for perms-granted events, but that can happen later, and is less
valueable while SMS is the only "dangerous" permission and the only one
that blocks message sends.
JSONObject(<string>, null) clears out any existing entry but doesn't add
a new one. So store missing names as "" instead. Not sure how this
worked when I first tested it....
My old asking for permission to turn on/off SMS is confusing and adds no
value when the OS separates out permissions and is confusing. So get rid
of it on Marshmallow and later, always returning true from the old API
and hiding the preference in that case.
The write-red-on-it thing doesn't work for the notify.png file used in
notifications (I think because I can't find a color Android doesn't
strip out.) So use a rotation transform instead. Users will never see
this anyway.
Move connection options READ_PHONE_STATE check from ConnViaViewLayout to
GameConfigDelegate and give it a rationale. Preferences stuff can't do
permissions because it can't override onRequestPermissionsResult (base
activity is of wrong class). Should fix that at some point, either by
moving to preference fragments or with a dummy activity to go over the
top.
Explicit "ask again" and "skip" buttons in alert showing rationale;
check/ask for permission before sending invitation via SMS; warn each
time SMS-enabled game is opened without permission but allow it to stay
open and if necessary send an invitation.
add "close game" button to warning about lack of comms, and do that on
dismiss too. Nothing good can come from having the thing stay open.
Eventually the "close" should turn into "edit" and launch GameConfig on
the game with the comms selector up, but that's hard, as currently the
launch only happens from GamesList and is via
startActivityForResult. Might be easiest to close and send an intent to
GamesList to cause it to launch GameConfig that way. Not for this release.
An edge case, but: doing "new from" on a game without any connection
types crashed because of an assertion in comms that assumed
addr_setType() was being called on zero-initialized flags, which
shouldn't have been a requirement. Pulled that as well as java code that
added RELAY-type connectivity to any game that had none. If a game has
none, leave it that way.
replace dlgButtonClicked() with separate methods for positive and
negative buttons and alert dismissal. I was making too many mistakes
because the old method was getting called twice (e.g. for negative and
then dismissal) and I hadn't needed to differentiate until adding that
new Action. There should be no behavior change with this, but it's
pervasive and replaces some spaghetti.
Let OS ask for STORAGE when user chooses Downloads dict storage option,
but do nothing if it's not granted. Existing code asks again when the
actual move is to done.
If an game's open that uses a wordlist kept in Downloads and permission
is revoked, close cleanly. On re-opening the user will be prompted by
existing code to download again. (Note: not prompted to turn STORAGE
permission back on. It would't be easy to detect that case, and I'm not
spending time helping people who are perverse.)
Add new static method that takes a request, rationale and Action, offers
the rationale if required and then, if user hasn't cancelled, asks for
permission. Eventually the Action is issued with either POSITIVE or
NEGATIVE. Use this to ask for STORAGE when downloading upgrades and
moving wordlists to the downloads area.
You need PHONE permission to start and play an SMS game but not to have
started it in response to an invitation. But to rematch you need it. So
ask, including offering a rationale that differs depending on whether
there's a way for the game to communicate once SMS is removed.
They were hacks for debugging SMS early on, probably infrequently
seen. Instead post messages. BUT: they aren't being received because of
how listeners get cleared in onPause(). Needs to be fixed, but there's
no harm in having them dangle for now.
Sometimes when a new game is created the board winds up with the toolbar
drawn over the tray. That's because when it was laid out the game wasn't
running yet and so none of the toolbar buttons was enabled, giving the
toolbar a height of 0. So now the first time the updated toolbar reports
that at least one button is enabled we force the board to lay itself out
again.
Little hack to dismiss a game when the invitations dialog was cancelled
broke invitations altogether. Revert that change, and simply check on
getting focus whether all players are present; if they're not, put up
the "invite or wait" dialog. This is a big change, but in 10 minutes of
testing I can't get the wrong thing to happen.
Remove strings that no longer exist in English, add missing "other"
quantities in Japanese (taking a guess from context that it's ok to use
generic counters), and in a couple of languages fix typos in format
specifiers that would have caused crashes.
It wasn't even checking all languages. Now it does, lets you specify
which ones if a subset's desired, and differentiates between sets of
format specifiers mismatching in a way that'll cause a
crash (e.g. expecting int in one and string in another) and just having
some missing.
Let's make it easier to add to the set of available BT devices from the
place where the user sees the need. TODO: update the descriptive text,
but not this release since it'll break existing localizations.
Channel wasn't a thing back then. Catch the ClassNotFound exception and
set the same boolean as when permissions aren't available (in manifest),
which means nothing else tries to run.
Add cancelled notification to invite choices dialog. Look for that in
board delegate and close the game. Makes the inviting experience more
consistent, the goal being that you never look at a game that's missing
players without some intervening dialog preventing you from trying to
play.
Use an array of actual objects rather than indices, and add an equals()
method, so checkbox status can survive changes in the set of available
items and even be reproduced when the objects backing them change (as
happens after a new BT scan finishes.)
Control whether wifi direct is enabled from build.xml, a file that's
already different for the two variants. The point here is to do a
release with permissions changes but without having to fix everything
wrong with wifi direct.
If user hasn't granted READ_CONTACTS permission, use the pairing of
numbers/names that's come from Contacts via the SMS invitation dialog.
Sometimes it'll have names for numbers that arrived via invitations. Do
some cleanup while at it to prevent stored BT mac addresses from being
looked up as phone numbers.
On launching SMS invite dialog, ask for READ_CONTACTS permission. Where
it's needed, though, is for the user-invisible process of translating
phone numbers to names (not launching the Contacts app via an
Intent). This just seems like the best place to ask since the user's
thinking about contacts and phone numbers.
InviteDelegate superclass now tracks whether something is checked,
though it's not doing as well as subclasses did before, e.g. when SMS
would uncheck everything but a newly added contact.
Add post-open test for SMS and then ask for SMS permission, giving
rationale if the OS says I should. Meant adding a new interface for
rationales and wiring that in. It's a bit spaghetti and I may think of a
cleaner way later.
When opening a game that sends via SMS, check if it has permission and
offer to remove the config if it doesn't, removing a similar check from
pre-opening code. (Here it'll catch e.g. games opened from the
GameConfig process.) Sends/receives without SMS permission are already
caught by existing catch{} blocks, so there's no crash.
Moving stuff into superclass and trying in general to simplify
things. BT and SMS invitation work, but the option to have more than one
player per device isn't available. RelayID invitation is a mess, but was
before I think. WiDir invitation crashes in assert because I can't
debug it on this branch. Lots of work and cleanup still required.
rewrite perms23 class to use Builder pattern, making it easier to pass
more than one required permission. Use that (with hard-coded set for
now) to check and ask for permissions a game needs to communicate. Offer
to remove the comms methods the user doesn't want to permit (but that's
not implemented yet.)
First, needed to remove some state caching since though a non-SMS
phone will never become one my ability to ask changes over time. Then
changed population of communication options to first ask for the
permission needed in order to find out what's supported, then to find
out what's supported via methods that will return false without
permission. Can't think of a better way right now....
Add a compile-time option to send commands in json as strings rather
than ordinals (ints) for easier reading of logs. And log/keep track of
when discovery's turned off or on in case it's useful for making all
this work.
In handling resend for a newly available transport, if the lock can't be
obtained see if that's because the game's currently open and if so post
to its thread.
Each time we get a socket to a peer, try sending all pending wifi direct
messages. It's a hack, but trying to send isn't that expensive and the
experience is much improved.
When contacting opponents, iterate over opponent addresses rather than
information cached in the summary. Start implementing for BT, but that
needs more work. Change enum names in WiDirService state machine. P2p
opponent notification untested because apparently an open hotspot, which
you can't prevent devices from connecting to, breaks P2P connectivity.
Rewrite to use new XWPacket class that wraps JSONObject but could later
support a binary format, and reply with new message when packet arrives
for a game that does not exist. Code already present turns that into an
invitation to delete the game.
fix race condition in accept thread; only call wifimgr.connect() on
guest side. Connectivity now works reliably if the guest starts, or is
restarted, when the group owner is running. If owner is restarted the
guest seems never to connect (without restart.)
Two crashes reported on the Google Play store in the 112 version have at
least easy work-arounds: catching exceptions and doing nothing. In both
cases that should be harmless, so assuming the causes are rare it's a
good move.
Make it a trivial state machine that implements Runnable and
ActionListener so it can be scheduled and notified easily. Removes a
bunch of code. Connecting still happens unreliably, but it should be
easier to tweak things now.
Add UI, minor thanks to recent refactoring, to allow invitations via
wifi direct. Uses a mapping of all currently known device mac addresses
to names, with only the latter shown to users. Works well, though
something I changed seems to have causes devices to start losing track
of their connections to each other.
Simple fix to the problem that a group owner may not be in the role of
game host: if I'm a guest and don't have a socket for the mac address to
which I need to send, slap src and dest addresses into the message and
send it to the group owner. That device can then forward it.
Found a recipe for service discovery that at least in the bit of testing
I've done is much more reliable. Devices connect almost
immediately. Basically everything chains through onSuccess() handlers.
URL to recipe is in code.
Examples kill it before calling connect(), so do the same. Seems to make
connection more reliable. (Pending: handle case where more than one
two devices are in the group!!!)
Made WifiDirectService into an actual service in order to better process
incoming packets. Now works for game messages from comms, and posts
notifications when app's in background. The latter requires using the
app context from XWApp since unlike the other transports this one
doesn't involve the OS invoking us with a Context.
Got to the point where I have an open socket for the packet I want to
send! Lots of changes to how discovery works seem to have improved
reliability though I'm still unimpressed. Much to learn.
Grab and store the local device's mac address. Add p2p as a type of
address, represented by the mac address of the recipient. Include the
local device's address in invitations sent when specified by user. Now
the WifiDirectService class is being passed a packet and the address of
the recipient; it will next need to set up sockets with every device it
encounters and map them to their mac addresses so that it can do a send.
Starting with the release of Nougat there have been cases where the
board would not correctly redraw. It's been most evident when using the
hit feature many times in a row, with something going wrong every fifth
time or so (but not that consistently.) It's as if modifications of the
bitmap backing BoardCanvas were being done asynchronously and not
necessarily all completed when I blit the canvas to the screen via
canvas.drawBitmap(). (As evidence of this I confirmed that a tap on a
tile in the tray after a bad draw would cause the screen to correct
itself even though the only additional rendering was to the tray. So by
the time that second drawBitmap() call happened the bitmap data was
correct: all draws that hadn't completed earlier had done so by now.)
The fix is to call Bitmap.createBitmap(bitmap) and to use the copy as
the source of the canvas.drawBitmap() blit operation. I suspect that
createBitmap() waits for any pending draws into its source to complete
before making the copy. Regardless, if this hasn't fixed the problem
it's made it so rare that I'm not seeing it, and since I'm only doing
the copy on Nougat there's little risk in the change. And I can't detect
any problems coming from the considerable additional memory being used
and immediately marked available for gc.
Starting with the release of Nougat there have been cases where the
board would not correctly redraw. It's been most evident when using the
hit feature many times in a row, with something going wrong every fifth
time or so (but not that consistently.) It's as if modifications of the
bitmap backing BoardCanvas were being done asynchronously and not
necessarily all completed when I blit the canvas to the screen via
canvas.drawBitmap(). (As evidence of this I confirmed that a tap on a
tile in the tray after a bad draw would cause the screen to correct
itself even though the only additional rendering was to the tray. So by
the time that second drawBitmap() call happened the bitmap data was
correct: all draws that hadn't completed earlier had done so by now.)
The fix is to call Bitmap.createBitmap(bitmap) and to use the copy as
the source of the canvas.drawBitmap() blit operation. I suspect that
createBitmap() waits for any pending draws into its source to complete
before making the copy. Regardless, if this hasn't fixed the problem
it's made it so rare that I'm not seeing it, and since I'm only doing
the copy on Nougat there's little risk in the change. And I can't detect
any problems coming from the considerable additional memory being used
and immediately marked available for gc.
commented-out logging of drawCell, with flags; debug-build checks that
static rects passed to java draw code aren't being used by multiple
threads at once.
Current code is hitting the relay every 60 seconds, at least on
non-google-play installs like f-droid. Switch to using a backoff timer
that maxes out at once per hour. Eventually may want to not run the
timer at all when there aren't any unfinished relay games present.
Now sends and receives a single packet more often than not. Disabled for
non-DEBUG devices, and new permissions that are required are commented
out so I don't accidentally ship before the sdk-23 permissions stuff
makes additions less scary.
Disabled at compile time, and missing permissions required, but if
enabled is able to discover other devices running same app. Next: try to
open a socket.
Meant using NotificationCompat, dropping FloatMath, and changing a bunch
of build config stuff. Not done for gradle builds yet. Currently crashes
on a new install until you go into Apps/Crosswords/Permissions and turn
on the four "dangerous" ones it's using. Now the work is to check for
and request missing permissions on demand.
Crash occurs when creating new game: because there's no turn set yet the
jboolean has no set value (whatever was on the stack). Apparently the
java runtime accepts only 1 and 0 for jbooleans.
Crash occurs when creating new game: because there's no turn set yet the
jboolean has no set value (whatever was on the stack). Apparently the
java runtime accepts only 1 and 0 for jbooleans.
This completes moving to using traditional logging. At least to having
each file provide a TAG. Class rather than TAG is passed in, and format
strings are preferred to contatenation, but now adb can work with the
output.
There are some screen dimensions, especially with dual-pane mode, where
the board is just bit narrower than the screen. Rather than have narrow
white borders, allow the cells to take up the slack. The API takes an
upper bound on the ratio of width to height so things shouldn't get too
odd looking.
Running into a case where views haven't been added to the layout yet but
I'm searching there for their associated fragments. Fix to NPE stops a
crash, but intents still aren't being handled. I need to try to handle
them later, but am unsure when to start trying.
Too-clever double loop had an no-exit path. Now it's simpler: find a
fragment that can handle the intent, pop all fragments to its right
until it becomes top-most (there may be none), and then give it the
intent. Use of popBackStackImmediate() is required so old fragments are
gone before intent handling wants to create new ones.
Since display is now by whose turn it is and how recently there's been a
move in the game, the order can change on every save. So in the listener
for saves, redraw the group the saved game belongs to. The best way to
do this without flicker seems to be to mimic closing then re-expanding
the group, so that's what I'm doing.
Getting an NPE iterating over fragment/children of dualpane view
because, I think, committing them's been postponed. So add the iteration
as a Runnable to the same postponement mechanism. This may be wrong:
needs to be run for a while to see if it fixes the NPEs I'm seeing (on
opening an intent in the morning after the device has been idle all
night.)
Attempt to fix crashes happening when the occasional fragment
transaction happens before the OS is ready. Keep a boolean indicating
whether we're in that state, and when it's not set create a Runnable to
be run next time it is. Temporary Toast lets me know when it's working.
Attempt to fix crashes happening when the occasional fragment
transaction happens before the OS is ready. Keep a boolean indicating
whether we're in that state, and when it's not set create a Runnable to
be run next time it is. Temporary Toast lets me know when it's working.
Problem: brain-dead Android code (on 5.1.0 anyway) caches the
SectionIndexer separately from the ListAdapter that implements it, so
when you change the ListAdapter it keeps using the old one for its
SectionIndexer. Solution is to replace the ListView in the layout with a
FrameLayout into which a ListView is inserted at runtime on init and
whenever the data changes.
