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.
Got a crash opening games where tile pick was on and first player was a
robot. Cause: non-robot player's tiles were being assigned before the
robot's, and the move stack didn't like the out-of-order assignments.
Now we assign tiles in order as before, but pause each time we find a
non-robot that needs to pick its own.
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.
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.
treatment of the final 8 bits. I'll need to release a version that
still produces the incorrect hash for compatibility with existing
clients that expect it, but that also looks for the correct hash. When
that's out there I can do a new release that sends the correct hash.
stack, go down the stack looking for a match. If it's found, pop down
to that point then apply the move. This fixes stalls in the test
engine when undo is enabled, and so I assume that undos and turns are
somehow coming in out-of-order. Should be rare that this happens.
can happen when somebody does an undo, since the ordering of undos and
moves is not assured. The result is a temporary (I think) stall, where
a move is rejected because the hash doesn't match, but that will be
accepted later when the undo's been processed and the hash will now
match.
# Please enter a commit message to explain why this merge is necessary,
# especially if it merges an updated upstream into a topic branch.
#
# Lines starting with '#' will be ignored, and an empty message aborts
# the commit.
side translate that into showing the sender's name in
notification. Not yet done: replacing silly "not me" in chat listing
with same, but now it should be easy.
Conflicts:
xwords4/android/XWords4/archive/R.java
side translate that into showing the sender's name in
notification. Not yet done: replacing silly "not me" in chat listing
with same, but now it should be easy.
Note: because the substitution is done in common code I can't use the
positional specifiers (%1$s vs %s) and so this breaks the generated
"translations". The scripts that do the generation need to be fixed to
understand the formatted="false" attribute.
quantity through to java world, use it, convert English <string>
resources to <plurals> (using python script) based on parallel changes
in French, and modify callsites to call getQuantityString() where
R.string.xxx became R.plurals.xxx.
inform of number of missing even when it hasn't dropped to 0 (so
second invite can be sent for 3+ device games); take down older alerts
before posting new (again since number of missing may have changed)
needs to save it there on receipt of an invitation (doesn't create
full game with comms until later.) Passes discon2 tests and seems to
work on a single Android device. (Haven't tested inviting on Android
yet.)
dict, give chance to switch, and to download if required. Because of
the way the JNI thread works, and JNI's requirement in general that
env instances match up (e.g. dicts must be destroyed in the same
thread that creates them), substituting into a live game is too hard.
So the game's saved with its new dict and then reloaded.
new param passed to remote and changing final score formatting and
menu items to match. Still need to show old menu on android in case
where game has already ended.
used by server. Clients need to care if e.g. the server's disallowing
phonies based on its dict. Can only be sent if client is of latest
version. In that case, common code calls into new util function. In
future changes, BoardActivity's implemention of the callback will need
to check if the server's choice of dict is available, and if not offer
to download it. Once it's available, will want to install it.
used by server. Clients need to care if e.g. the server's disallowing
phonies based on its dict. Can only be sent if client is of latest
version. In that case, common code calls into new util function. In
future changes, BoardActivity's implemention of the callback will need
to check if the server's choice of dict is available, and if not offer
to download it. Once it's available, will want to install it.
if all tiles are used (ending the game and triggering penalties for
other players.) Changes are protected by compile-time flags currently
not turned on.
-- and to detect when an incoming move doesn't make sense. These
latter changes may not be necessary now that hash code is checked
first thing, but can't hurt, and there will be devices without hash
codes for a while.
when a devices sends to self via SMS and client winds up getting a
registration message it just sent. Not perfect yet (and may require a
change in the comms protocol -- that's not versioned :-(.
so that snapshotting made score strings disappear. Also rename
stream_copyFromStream to stream_getFromStream to make it clearer that
the src is modified. And swap params in stream_setPos().
it if any players are missing. Idea's to allow for non-relay devices
the invitation opportunity that comes when a game connects to the
relay and learns that no other games have joined its room.
with 15x15 and 17x17 (and presumably larger) with the old games being
upgraded mid-game: when server's new and wants to use a larger board
but not all clients are capable, shrink the board; and, when new
server is opening state saved by pre-upgrade code, assume a non-0
proto version for clients that registered earlier (as old code didn't
save the client's proto version.)
clients to append their stream version to their inital connect
message. (The format can't change, so detecting additional length was
the only option. comm.c on existing clients won't allow more than one
connect message per channel, so adding a new to be used in addition
didn't work.) New servers detect this; old will ignore. Track the
version (implicit or not) of all clients, and use the lowest any
supports, so that new server and all new clients will use newer proto.
MAX_COLS was larger than 16. In order that old-style messages on
relay be readable by new-style code on device, modified server.c's
protocol to include stream version. But: unless I come up with a
better way of doing this all devices will have to be upgraded at the
same time: old won't be able to read the new format as it's done here.
to enable it for a game yet). In common code, break pick tile utils
method in two, one for blank and the other for tray tiles. In java,
implement dialog for picking the tiles. Works, though is a bit clumsy
to have the dialog come and go after every pick. Better would be a
dialog that stays up and accumulates picks -- but it'd have to know to
reduce the set of tile choices as letters are chosen. This'll do for
now.
played that include that tile and pass to new util_cellSquareHeld()
method. In java implementation of that method, use existing lookup
activity code to display list of words. Enabled on the C side by a
compile-time flag in case it has problems. Right now the time spent
saving a game before launching the lookup activity, and reloading it
after, is pretty apparent, but that's in emulator which is slow.
util_userQuery(QUERY_ROBOT_MOVE) with new util_informMove(), into
which the number of words formed and the words themselves are passed.
The process of consing up the score explanation was already passing
over the model, so storing the words is very little effort, and will
save a call back into the model where the user actually wants to do
the lookup.
uses existing stack and undo features to run a WordNotifierInfo over
some number of scoring passes to gather the words seen. Seems to work
as tested from gtk version. Now need to try from android....
one and changing the site that used it to use the other. Not visible
outside of common, but should make it easier to harvest a list of all
words for one move or the entire game.
where games with more than two devices would hang because server.c
code was dropping messages that comms.c code thought were good and so
ACK'd preventing them from being sent again. They were being dropped
because the game was in the wrong state after displaying a move-made
dialog because the state it was to move to after doing that display
had not been saved.
passing a 0-length array of bad words. That in turn suggests a
screwup where a move's rejected for some reason other that a word not
being in the dictionary. This is all supposition, and all stuff
that'd be caught by asserts in a debug build, but: when there are no
bad words don't report them -- even if the move's rejected.
not yet received on device. Meant to be included in summary. I'm not
happy with how much code it took to figure this. I don't know
server.c all that well any more.
player. Works for gtk client. Compiles for Android but there's no UI
yet to specify more than one dict. Management of dupicate dicts
without duplicating memory -- refcounting -- will be up to the
platforms.
implemented (when not smart) as trying to match the human's score to a
per-robot value between 1 and 100 that gives the percentage of best
moves to store before picking randomly from among them. So a 1 means
save only the best move and always pick it; 100 means save all the
best moves (how many are saved is compile-time configurable) and pick
one of them. Because it's settable per-robot a smarter robot can be
played against a dumber one (though I may choose not to make it
settable per-robot on shipping versions.)
into the java world by making it a util_ctxt function. Do same on
linux to test. We'll see how it is -- and can back this commit out if
there's no improvement.
meaning it'll work on any tranport but relay doesn't know about it.
(If relay knew about it I could send a "shutting down for a minute"
message to every connected device, which would be cool. But this is
easier.) Written, BTW, enroute from Seattle with help from beer from
Chris in first class. :-)
