The only significant change is moving CustomViewTargetTest into the
library project and out of the test project. This appears to be
necessary to work around issues with transitive resource dependencies
across libraries. Hopefully we can find a better fix in the future.

ShadowDrawable.equals() is being removed in cases where possible I migrated to ShadowDrawable.getCreatedFromResourceId(). In other cases this was not possible and the existing equality check was returning a false positive because the comparison, intrinsic height, width and bounds were all uninitialized.

PSA: [] for unit tests that are ~25% faster, higher fidelity + simpler configuration

-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=206187349

1) Synchronize only the inner method that accesses the cache, as the outer call
that loops and filters the loaders appears to not have threading issues.
2) Probably less of an impact since I imagine the loops are generally small,
but only allocate an ArrayList once you find a matching loader, and allocate it
with size - i which is the max possible remaining in the loop, rather than
always allocating it to size.

-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=205877123

[third_party] Avoid saving Robolectric shadow instances; instead, save the real framework object, and get its shadow when needed using Shadows.shadowOf(). Call Android framework methods directly on the framework object instead of its shadow whenever possible.

LSC: []
Additional details: []

Cleanup change automatically generated by javacflume/refactory
Refactoring: //third_party/java_src/robolectric/errorprone:ShadowUsageCheck

Tested:
    TAP --sample for global presubmit queue
    []

-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=204784445

The Android Test Support Library is being renamed to AndroidX Test, with its root java namespace changing from android.support.test to androidx.test.
And code location changing from tp/android/android_test_support to tp/android/androidx_test.

More information: []

TAP global presubmit: []

BUG:
79205333

-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=202796842

The two main differences are:
- It forces you to override the method where resources must be cleared. Not doing so results in recycled bitmaps being used and crashing apps. Not doing so was a common pattern among developers optimizing for lines of code instead of correctness.
- No more setTag(object) use. Glide now targets 14+ which can safely use the id tag variant and avoid another class of runtime bugs caused by developers optimizing for lines of code instead of correctness by calling setTag() and overwriting Glide's data.

Finally, we deprecate ViewTarget, SimpleTarget and BaseTarget. Apps should primarily be using Target, CustomViewTarget, ImageViewTarget and FutureTarget which either force the developer to implement all necessary methods, properly implement them themselves, or will not attempt to reclaim bitmaps. The deprecated classes continue to be used internally by some of the "correct" classes but can be merged down once the deprecated APIs are able to be removed.

-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=201729878

Math.round() called with an integer or long type can result in loss of information because it coerces the argument to a float before rounding back to an int.

If this CL introduces an explicit saturated cast from long to int, you may want to remove the cast if it wasn't desired. If so, please make a suggestion to remove the cast in this CL.

NOTE: Review carefully. If the original code contained the pattern Math.round(a/b), there's a good chance the original author wanted a floating-point division with round-to-nearest instead of the integer division with round-to-zero that they got.

This an automated fix using Error Prone ([] We intend to make this a compilation error in Blaze's Java compiler, so it is necessary to fix existing occurrences first to avoid breaking builds.

More information: ​[]

Tested:
    TAP --sample for global presubmit queue
    []

-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=201592824

Various other changes allowing Requests to restart have removed the
assertion that cleared requests are never restarted. There doesn't seem
to be any need currently to distinguish between a request that's cleared
by a call to Glide.with().clear() and a request that's cleared by Glide.
We can simplify the API and reduce confusion by just calling clear()
directly and removing pause() entirely.

-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=201439855

It's no longer necessary, the Target is always cleared first whenever a completed load might be present.

-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=199811222

*** Reason for rollback ***

b/80539264

*** Original change description ***

Default allow hardware config to true in Glide for P+

For more information on hardware Bitmaps, see []

***

-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=198772333

For more information on hardware Bitmaps, see []

-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=198730947

The name is only used for in memory cache keys so changing it is safe.

-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=198598261

-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=198569187

Pausing the request allows Glide to show a placeholder without starting the request.

*** Reason for rollback ***

Rollforward with fix for b/80259796

*** Original change description ***

Automated g4 rollback of changelist 197922991.

*** Reason for rollback ***

b/80259796

*** Original change description ***

Default hardware bitmaps to off via the Downsampler option.

Right now we default hardware Bitmaps to off via a deprecated DecodeFormat enum value. We want to remove the deprecated DecodeFormat. Moving the default off state for hardware Bitmaps into the non-deprecated option will let us complete that migration without also defaulting hardware bitmaps to on.

Hopefully in...

***

-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=198089173

Glide will always check for a placeholder/error/fallback drawable first
and only check the id equivalents if the drawable is null. Previously,
when setting a placeholder/errror/fallback id on a RequestOptiosn
object, we did not unset the drawable equivalents. As a result, it
wasn't previously possible to replace a placeholder/error/fallback
drawable by setting the equivalent id.

After this change any calls to placeholder/error/fallback will always
replace previous calls, regardless of whether and id or drawable is
provided.

Fixes #3068.

-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=195990546

serves from https now

-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=195673007

Fixes #3061.

Related to #2894.

-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=193438586

*** Reason for rollback ***

b/78110139

*** Original change description ***

Use ExifInterface to parse orientation data on OMR1+ in Glide.

***

-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=193431565

-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=192690571

-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=191378721