| Commit message (Collapse) | Author | Age | Files | Lines |
... | |
| |
| |
| |
| | |
Change-Id: I17d4eae699e55e41c26c6125175b893a3012699e
|
|/
|
|
| |
Change-Id: Iad1576ec4ca03213679aabb97328bcf9c0f590eb
|
|
|
|
|
|
|
|
| |
It can't be used at the same time as the new one
(the SDK_TEST_BASE_URL one) as it is confusing if
both are defined.
Change-Id: I5b3ab7bac5d58ab9b1e47e2aea43229156c30083
|
|
|
|
|
|
| |
SDK Bug: 5508174
Change-Id: I14ab4163e0f07824226cc78f72b538e2f82d7c5f
|
|
|
|
|
|
| |
SDK Bug: 5508174
Change-Id: I466ba63c7f6c4158be42df1151f540d26c63e342
|
|
|
|
| |
Change-Id: I15d0b46ef09791ea000724e571de63923f147b6d
|
|\ |
|
| |
| |
| |
| |
| |
| |
| |
| | |
SDK Bug 20535.
Doesn't seem reproducible, so just defend against
a potential NPE anyway.
Change-Id: Ieb6a0958c7585ea3c5cffbb640943263a14e4c60
|
| |
| |
| |
| |
| |
| | |
http://code.google.com/p/android/issues/detail?id=21006
Change-Id: Ice55f026e9d78922bfa359edad62630a8f7fb8ed
|
|\ \ |
|
| | |
| | |
| | |
| | |
| | |
| | | |
SDK Bug: 20963
Change-Id: I46b8e808f2eb0c008b60036631a4d061d02085c6
|
|/ /
| |
| |
| |
| |
| |
| |
| |
| | |
- Clearly states that a package is not compatible.
- Provides an env var to work around and install anyway:
$ export ANDROID_SDK_IGNORE_COMPAT=1
$ tools/android
Change-Id: I5d94fbe452c72572054b2aecb781acfe3519135c
|
|\ \
| |/ |
|
| |
| |
| |
| |
| |
| | |
Should have been in a3e43f30cf2278afd5dbc5864442529129b3a057
Change-Id: I9018f8b3b0e5bfb07b05acba9d1c6d4fae04181a
|
|/
|
|
|
|
|
|
|
|
| |
SDK Bug: 20939
Incidentally, this also fixes the fact that the window
for the SDK/AVD manager can't be resized at all on Windows
when started from ADT. SWT is quite magical.
Change-Id: I282ebdac3e42e39ee41247de5591cfcb61cd4ac8
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This changes the way packages are selected:
- When using the "Select New" link, all new stuff is selected.
- When using the "Select Update" link, only installed stuff
with updates is selected.
- When starts the SDK Manager, the heuristic becomes:
- Select any updates we can find.
- Check whether the top platform is installed.
- If not at all, select all its packages. This should
cover the "there's a new platform available" scenario
by just selecting to install everything for it.
- If the top platform has at least one item selected,
make sure the platform package itself is installed or
select it.
- If the platform currently installed or a selected
update lacks a system image, we want to suggest some
kind of system image. For that, look whether the platform
(current or selected update) provides one. If they don't
select any system image package we can find.
- On Windows, also suggest to install the USB driver.
SDK Bug: 20607
Change-Id: Ifab423b226ad708e9117eefd4d76033d866c57a8
|
|\ |
|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Support a new env var SDK_TEST_BASE_URL that should
be set to the root of an alternate repo. When set
it will completly override the built-in default
repository base URL.
Change-Id: I4852d4c60aac19e2e5c7c34d848b15254d93afb6
|
|\ \
| |/
|/| |
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
The BuildConfig class contains a single boolean constant
called DEBUG which is true only for debug/instrumented builds.
This allows developers to create debug only code that's automatically
stripped from release builds.
Also fixed some issues with dependency check for other tasks, notably
the aapt task that would always find new files due to not filtering
out files that are ignored by aapt itself (hidden files, version control
files, etc...)
Change-Id: I4391a87c064a185d6b337ca46e3a9f0e43c5174d
|
|\ \ |
|
| | |
| | |
| | |
| | | |
Change-Id: I0518515a2095b63099cb69d67d110330b148d5a2
|
|/ /
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
This is exactly like change I6f82b040 except now we
remove the stderr output when parsing the addons_list.xml
To put this in context, end-users should not see any
xml parsing error. In this case we get one because the
file doesn't exist on the test server and the web server
returns an HTML page describing the error, which naturally
fails to validate against our XSD. But since the point of
the method is just to validate, errors are irrelevant.
Change-Id: I80b2c7600d1ceffa870ac3d71ef0c0f6ef9f809c
|
| |
| |
| |
| | |
Change-Id: I52484e7b61100d0c1656769cdce419416ed060c6
|
|/
|
|
|
|
|
| |
This was due to the shell being marked as app_modal.
SDK Bug: 20616
Change-Id: I1352fe4573f79890f2633be6140c1c63b79150b6
|
|
|
|
|
|
|
|
|
|
|
| |
Have APPLICATION_MODAL causes the log window to never have focus.
This patch has been tested only on Linux.
Updated by Raphael:
- Differentiate behavior for Linux.
- Also apply to the AVD manager standalone window.
Change-Id: Iff8461e26c3e36712083708969a781b9d5ed63c8
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
SDK Manager hangs on windows after it tries to stop
and restart ADB (e.g. when installing platform-tools).
It hangs when capturing the stdout/stderr pipes -- these
don't close automatically when the process has finished
and these are not interruptible streams.
One workaround is to not capture them, especially since
the output isn't really useful for the installer anyway.
Change-Id: I6554461dfffad2cc8ff0f1fe7d212fdee742e2e6
|
|
|
|
| |
Change-Id: I24b49b4b09a9cab8cb1ddd2eade78e9ea21cd867
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Purely an UI issue, the Install button is not enabled
correctly for updates, thus preventing users from updating.
Patch 2: changed to have the same code used for both
enabling the buttons and actually selecting what to install
or delete. That will avoid inconsistencies.
Also prevented against a few potential NPEs.
Change-Id: I3d209fa4325ffc7624c503395834636b372e8e7e
|
|\ |
|
| |
| |
| |
| | |
Change-Id: I6f82b040e8a2f91c24b7e75b3371f249b0b0f383
|
|/
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This splits the "Select New/Updates" link in the
SDK Manager in 2 parts. Behavior is:
- at startup, the sdk manager selects both new and updates.
- one can click on deselect + select new to select
just new packages.
- one can click on deselect + select updates to select
just update packages.
- select does not select first, so clicking both new
then update will produce the previous behavior.
SDK Bug: 20607
Change-Id: Ib861d78b746f4f3c7a8fce421c89ae8094339933
|
|
|
|
|
|
| |
See http://code.google.com/p/android/issues/detail?id=20447
Change-Id: I7db8f16b4027a9f4582c4a70bf8ad1c048ae7006
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This changes how source URLs are handled. Packages
from different sources were previously treated as
separate if the packages were the same (e.g. same
platform API) but the source URLs were different.
Instead this checks the hostname+domain name of the URL
is different, as well as the type of the source (that
is a sdk repository vs addon repository).
Change-Id: I5ba133cd0dc984528bf696aef77d00cd9d31ed9c
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
In various parts of the code base, I see code that creates
a FileInputStrea and never closes it. Unfortunately on
Windows that means the file will stay locked for as long
as the java process is alive.
In this case the SdkManager wasn't closing to input streams
which makes the source.properties files locked, and thus
it makes it impossible to upgrade/delete the corresponding
packages.
There are more occurences of this elsewhere in the code.
Change-Id: I9f0a854b210411757eee8b4ebf8308b059206542
|
|
|
|
|
|
|
|
|
|
| |
Properly enforce that all local archive should have
an actual installation path to be considered local.
This was missing in the system image and the source
archives introduced last week.
Change-Id: I3e4abd3c4136c9be73061e207e42e2d3cc22ca8b
|
|\ |
|
| |
| |
| |
| | |
Change-Id: I715a7503a7be2b28cd89bd441b8cbb5ee620ccac
|
|\ \ |
|
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
(This is part of Change I566b6945 but I had forgotten
to amend & upload, grrr.)
Change-Id: Ic494179481346a2eda0f56327ddac0f123bf4538
|
|\ \ \
| |/ / |
|
| |/
| |
| |
| | |
Change-Id: I566b6945b7d31c3bfe52b834014beb3f37a098d8
|
|/
|
|
| |
Change-Id: I4affda9be00c37c4a801816625f57f7eda585ef4
|
|
|
|
|
|
|
|
|
|
|
| |
This fixes the bug in the current code.
We're going to switch to a different upgrade mechanism,
rendering this code useless, but I'd still have this
checked in for the record: the issue is that the
temporary node is mixed up and the attributes use the
element name instead of their attribute names.
Change-Id: I3abcfd476494ee2ddb8b75789b346e86556d8d85
|
|
|
|
|
|
|
|
| |
When SDK_TEST_URLS is set, it prevents the default
sources from being loaded. This makes it easier to
test just a new site.
Change-Id: I463759ca43c7836e81524c2eac4dad769bc7032c
|
|\ |
|
| |
| |
| |
| |
| |
| | |
SDK Bug: 20430
Change-Id: Id427390c799165a1dff07530c6fce48488193055
|
|/
|
|
| |
Change-Id: Ib8df984ddbad0bb6f0eb4c2fb0bd4b9c49b334cc
|
|
|
|
| |
Change-Id: Id9cacdc53e2e800526950480007e0a8f474779be
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Fixes for the first time install support:
- Support installing more than one platform.
The custom install task was tailored before to one
install a single package and return one status with
on archive install path. This has been made more
generic.
- Fix a couple occurences of the dreaded "SWT Shell disposed"
error that happen when updating the progress bar and closing
the window at the same time. The issue is that I was checking
isDisposed before starting a syncExec but the window can be
closed between the moment the syncExec is schedule and when it
happens so code must check again for isDisposed inside the
syncExec.
Change-Id: I413cfacd63406febc7953effad5ece2ef2854360
|