…no such device" on second & subsequent accessory connections. Sub-optimal but "works".
…process. Start of an attempt to fix the "no such device" error.
…t. An attempt to solve the "no such device" error. Unsuccessful but keeping it anyway.
…r. Didn't work. Should be redundant (because closing the file descriptor parcel should close the underlying streams) but leaving in anyway.
…sue (mostly unsuccessfully).
… communications methods. The end result of a *whole* lot of stuffing around trying to get things to work. While a launcher screen was always intended, its addition was driven in this case by a need to make up for the flags "FLAG_ACTIVITY_TASK_ON_HOME" and "FLAG_ACTIVITY_CLEAR_TASK" not being available for Gingerbread. Note: The previous code did work for non-Gingerbread situations but was a little bit less than ideal. Adding the launcher screen necessitated a whole lot of stuffing around to try and get all permutations of not-yet-started, already-started, already-run-once, on-network-screen and usb-auto-launch-occurred working correctly. Which they now do--*but* only on Honeycomb and above. On Gingerbread a USB connection only works once. The second time we still incur a "device not found" error when the handshake data is sent. (But this AFAIK is not due to adding the launcher.) I'm committing this work now since it at least preserves the existing correct behaviour on Honeycomb+ and I'll look at debugging the other issue on Gingerbread separately.
…an throwing an exception. Note: The comment probably isn't going to be true anymore with the next commits but it's probably better having the check there rather than just barfing if the "impossible" happens.
…ve both the Network and USB implementation in one library folder. (See header files for further explanation.) Prior to this change you needed to attempt to compile a USB demo sketch and a Network demo sketch before compilation would succeed. For me it's more important to have only a single library install no matter what Handbag communication protocol is used than to avoid this hacky work around. --HG-- rename : library/Handbag2/Handbag.cpp => library/Handbag2/Handbag.cpp_ rename : library/Handbag2/NetworkHandbag.cpp => library/Handbag2/NetworkHandbag.cpp_
--HG-- rename : app/res/drawable-hdpi/ic_launcher_handbag.png => app2/res/drawable-hdpi/ic_launcher_handbag.png rename : app/res/drawable-mdpi/ic_launcher_handbag.png => app2/res/drawable-mdpi/ic_launcher_handbag.png
… by the match method.
…ard! w00t! :) Note: Still no auto-handling of disconnect. Note: 2.3.4 seems much less stable than 4.1--reconnects without resets don't work; also, second connect causes network screen to open rather than usb, and it seems to sometimes get stuck expecting a USB device even in local network demo/resume mode. Note, also: In 4.1 a long press is required to process text input. Also, also, it seems like SMS is back working again--although not 100% (although that might be just the bad cell service). --HG-- rename : library/Handbag2/NetworkHandbag.cpp => library/Handbag2/Handbag.cpp rename : library/Handbag2/NetworkHandbag.h => library/Handbag2/Handbag.h
…SB source--and in general has been a total PITA. Especially since I'm not really using it correctly mostly. A hacky change to make it work to meet a deadline. Should really be done "properly".
…from within a service--like we're about to do. (Note: This breaks the code with this commit.)
…s the only activity seen. i.e. When on the display screen, pressing the back button exits the application.
…setup :) ) with a one packet hardcoded response for demonstration purposes. Note: Currently we only setup the connection once the parse server has been connected--this may not be necessary once we've improved the implementation. Currently this doesn't handle the back button from the display screen in a useful way.
…ice we can't start the activity until then. Thus we now start the activity only after the service is bound. (If we don't do this then we end up passing null as the comms service--I'm not sure why we dont encounter the same problem with the network comms service.)