Due to public ID assignment, gaps in resource type and entry
IDs can occur. This can lead to null pointer dereferencing
if not careful. This happened in ResourceTable::getDensityVaryingResources()
which is called when building Splits.
Bug:30879690
Change-Id: I588e4dcd2e042fccfcb2e87967b5cbd0d23b4497
(cherry picked from commit 081d1b4cf602fdd7302b597e6bf902cb415bc3a8)
Doing so is likely to trigger log messages in the future, and will
hinder diagnosis of other problems.
Bug: 30292538
Change-Id: I2cb8c507e43d06d4afa7e35b564416c1474c81d1
ID resources that get generated on demand with the
notation @+id/name were previously not given the
appropriate type ID offset when being built as feature
splits.
This change declares an ID type ahead of time so that
the type ID offset is applied before IDs can be generated.
Bug:30607637
Change-Id: I122a9133cb01b35e9892103ec52fc228dc65bf1a
This changes the default implied feature of 'android.hardware.touchscreen'
to 'android.hardware.faketouch' if no 'android.hardware.touchscreen'
feature is requested, required or otherwise.
Bug:30571641
Change-Id: I1e41242d4b1dc549cf69741d2a309baf476d084e
AAPT will continue ahead without reporting an error if a file
failed to be added to the ResourceTable. This would cause crashes
later when the file was assumed to be present.
Bug:30200166
Change-Id: Ieb2daf97ccf0345153b6f4598d130a38d108c937
Previously, AAPT would match locales even if they had different
explicit scripts if the requested locale was just a languages. Now
we require explicit listing of the languages with different explicit
scripts in order for the locale to be included.
Bug: 29412034
Change-Id: Ia118a5a7f9aec49f6c3c53b9195a0ae1a57f53fd
We need the attributes to remain public because people might still be
linking against them, but we don't want them showing up in the
documentation any more. Them showing up in the documentation also had
the side effect that it would accidentally mark the parent class of
attributes as @removed, which was not intended.
Bug: 28663748
Change-Id: I2f6eb09455fddf1086e6b24bc3bea5292e8e32b7
Create an analogue of "aapt -G" which outputs a proguard configuration
that keeps only components which need to be in the main dex.
Bug: 27383099
Bug: 28425556
Change-Id: Ic18c8c563794ff27a5598a214111d1b446a005f1
(cherry picked from commit 86229cb622fccde8ab8cbe85eead91a34313a708)
Create an analogue of "aapt -G" which outputs a proguard configuration
that keeps only components which need to be in the main dex.
BUG: 27383099
Change-Id: Ic18c8c563794ff27a5598a214111d1b446a005f1
(cherry picked from commit 86229cb622fccde8ab8cbe85eead91a34313a708)
On certain devices running Android M with unknown vendor
modifications. Setting the localeScriptWasComputed bit in
the resource parameters struct causes the translations
for many languages to fail to load. The app then
defaults to English.
This has been reported on the following devices:
Motorola XT1086, XT1096, and XT1586, HTC One, and Huawei
MediaPad 10 LINK.
This change works around the issue by never setting
localeScriptWasComputed in aapt and instead always
writing the default value of false.
Bug: 27872476
Change-Id: Id638d1d0e5f9dcb1c02e65d67667d76174c69b9e
Because there is a bug decoding paletted images in SDK_JELLYBEAN,
we need to avoid encoding paletted images for apps that support
SDK_JELLYBEAN and earlier.
BUG=27643907
Change-Id: Ib7d51ed87435cd36507915d62b0057c06f18b2b6
Previously, a bit was kept to find if the script of a locale was
explicitly "provided" in a resource. This was not backward
compatible, and failed in some edge cases when the package was
created with older versions of AAPT that did not set the bit.
The cases would happen when the old resource had an explicit script
specified in its locale, but since the "provided" bit was not set in
the package, we would assume that the script was computed by us.
This CL replaces the "provided" bit with a "computed" bit, so the
default value of the bit (set to "false" for old packages) would be
correct.
Bug: 27156990
Change-Id: I99e7f1ad8f70c90e25ab3640ed34cc1a6f8d1d64
This allows libraries to coexist a bit easier, since a lot of them
declare the same attribute. When used together they cause errors.
Bug:27101533
Change-Id: I649cf17bc587c4814f849d5ac522daf74495adcf
This allows libraries to coexist a bit easier, since a lot of them
declare the same attribute. When used together they cause errors.
Bug:27101533
Change-Id: I10327251636392e810637a79e7c9aa514f3f07a6
There are so many public resources that do not warrant comments
that having a warning that no one cares about is just producing
log spam.
Bug:21495023
Change-Id: I84e3b39c7139ee67fb5cb711d5ab466d8f4ef2e0
Many PNGs have multiple distinct pixels with zero alpha. This
CL allows us to treat all of those pixels as a single zero pixel.
This saves space in the color palette and also may help save
memory pages when we skip writing zeros.
There is also a potential disadvantage: if someone intended to decode
this image as unpremultiplied with particular transfer modes where
their color components have an effect even though alpha is zero,
this will prevent that. At the moment, we do not support unpremul as
a source, so this won't affect anyone trying to draw with the view
system.
This change originated in:
https://googleplex-android-review.git.corp.google.com/#/c/854580/
Change-Id: I702c7bd22d431cc7c775ed29bbd73c930f945ca3
