The pngcrush tool does not add any noticeable value any more, such that it justifies the extra build time: 240MB vs 238MB.
The build.prop changes are now no ops, since the key for ro.kernel.android.checkjni no longer exists.
Changing the ro.build.type from eng to user is also a no op, since CM is now userdebug.
Deleting bins is also unncessary now.
Stripping modules is also unnecessary now. (plenty of space)
At this point, I think we can just symlink $OUTFILE to $OTAPACKAGE, and not even bother
with a copy.
Change-Id: I431ac6425dedb6cd02eb2d31fc9ff82b49df9f50
- add new file for common functions and stuff
- though it doesn't contain any function now ;)
- red color for errors
- yellow color for "Optimizing: *"
- cyan color for modversion :)
- green color for "package complete"
- and some general cleanup
Change-Id: I83b2e352580e99b1766c5e63821281224cde5c49
Instead of setting the OTAPACKAGE name here, expect
OTAPACKAGE will be passed to us. This fixes an issue when
BUILD_NUMBER is defined in an environment with automated
builds.
Change-Id: I0634416e5332790d314f6a70313ea67821fd1331
from: http://review.cyanogenmod.com/#/c/19170/
This change enhancement the optimization of cm ota package, allowing do the optimization of
framework-res.apk in a parallel way. The current script optimize first the
framework-res package and then the rest of packages with xargs in parallel. The framework-res
package can be processed with xargs in parallel copying all the packages to a temporally folder
to apply xargs (because opticharged need that the package is the current directory).
This shortens the time of optimization process, because while framework-res package is
optimized, the rest of processor cores can be optimizing other packages.
Change-Id: I047f8c01b066db99e4dcc8bcb2f30c72492e1af8
Now uses $OUT/../../../host instead of $ANDROID_BUILD_TOP/out/host
fixes build when used with OUT_DIR_COMMON_BASE
Change-Id: Ib1934b57f3e94a3c3af10189c803fc494438a4a4
* Remove "update-" prefix and "-signed" suffix.
* Replace SNAPSHOT with EXPERIMENTAL (arcee).
* If CM_EXTRAVERSION, force to type EXPERIMENTAL.
* UNOFFICIAL builds now with datestamp in the same manner as NIGHTLY.
* date -u uses UTC timezone for consistency.
* Remove some CM7 dead code.
Change-Id: I681fd92e27aaf62c50eae20265ce07c67d477bca
removed functions for working with a squashfs image if target was dream
removed a useless comment (deleting wifi module, it's device specific)
Change-Id: I3d634f572bc3bf975441dd562f6e8111c0e60764
I got "java.lang.OutOfMemoryError: Java heap space" error when I'm building own Kang (galaxys2 for example) with Gapps included.
Increase maximum heap size to 1024MB will solve it.
This avoids printing an error message and usage statement when using squisher if no modules
have been produced.
Change-Id: Iadd38518e64f7edf2d4da345143a44970c6f383a
Signed-off-by: Pat Thoyts <patthoyts@gmail.com>
Number of processors: use sysctl on darwin.
xargs: fix incompatible arguments.
sed: use gsed on darwin because -i option is not compatible.
Change-Id: Ic3885db592e9588f01b69843add0e287ad35d4e9
Before
real 1m25.583s
user 1m14.101s
sys 0m21.717s
After
real 0m47.257s
user 1m46.123s
sys 0m23.453s
Change-Id: I1ca0013f8dec9175027f2e9bed93786cfe3804d5
If mksquashfs is unavailable then set NO_SQUASHFS true so that a valid
OTA package is created.
Reduces the amount of output by (optionally) using quite zip/unzip
Make better use of sed (we don't need grep|sed as sed can match input).
Something isn't setting the path correctly when running in a 32-bit
chroot environment in ArchLinux. This causes the script to cd into
$WORK/xbin. The PATH incorrectly contains ".", so all attempted
commands (rm, mv, cp) fail to execute as it is trying to execute
$WORK/xbin/{rm, mv, cp}.
