Every time gdbclient.py finishes it's giving this error:
Exception TypeError: 'must be type, not None' in <bound method
_subprocess_Popen.__del__ of <adb.device._subprocess_Popen object at
0x7fdd48e76850>> ignored
Basically super(_subprocess_Popen, self) was failing because at that
point _subprocess_Popen is None since global teardown had already
deleted the reference.
This fix saves local references to required global objects so that they
stick around until we're done with them. We may want to start looking
into making _subprocess_Popen a context manager instead, usage won't be
as convenient but __del__() does not scale to general use well. For now
though this CL seems to solve the problem for gdbclient.py.
Change-Id: I6c2a9fe0c108ae600cf05eb3b887d56959c7308e
adb.shell() was recently changed to use subprocess.Popen(), which
doesn't work properly with Unicode on Windows. The fix is to use the
same work-around that I did for subprocess.check_output(): write UTF-8
to a batch file and run it. The change is primarily refactoring to
enable code reuse.
Change-Id: I88e9b9b35e5318533c0cd932d92e13bc9e734092
Signed-off-by: Spencer Low <CompareAndSwap@gmail.com>
Add a test to send stdin to non-interactive `adb shell`.
This new test will hang on ToT devices that support shell_v2 until they
are rebuilt with the corresponding adb CL. A new feature could be added
instead to filter those devices out, but it doesn't seem worth it as no
devices have yet been released with shell_v2.
This CL also fixes a mistake I made earlier with device.linesep; this
is used to check device output, not separate device input. Using it
added unnecessary newline characters on Windows.
Bug: http://b/24565284
Change-Id: Ic123402975033d74688f56a36acac993af6815e6
http://r.android.com/173810 changed the shell feature name. This CL
updates the python scripts to re-enable shell protocol tests, and adds
an additional test for `adb shell -x`.
Change-Id: I961adf7abd1782ea98f3f82d8af11dcaa02f9570
This is mostly to allow backgrounding of processes run through adb
(gdbserver in particular).
Bug: http://b/23715403
Change-Id: I47fbebbd05d58044b4c447ffa86e0ab97d920278
The install escaping test fails if the device supports exit codes. This
fixes the test to ignore the exit code since we expect the install to
fail.
Change-Id: I37bb39552a6e9068ac088a25d831f45d81b28155
Adds python tests to check that -T (disable PTY) and -t (force PTY)
arguments work as expected for `adb shell`.
Bug: http://b/23825231
Change-Id: I5343fae35b2be8459a9b95125f66def46c26adf4
This really ought to be maintained as a separate project. Also adds a
setup.py for distribution.
Bug: http://b/22881740
Change-Id: I8e11e4154fac21d3196a533613822b236b2b73e6
