BookmarksTest.testDownloadMovedBookmark() expects that sync returns newly created items in LIFO order.

https://cs.chromium.org/chromium/src/chrome/android/sync_shell/javatests/src/org/chromium/chrome/browser/sync/BookmarksTest.java?rcl=0&l=163

Enabling autofill profile de-duplication, which is not otherwise expected to interact with this test, causes the ordering expectation to no longer be satisfied on re-runs of the test on a given machine.

This behaviour is only observed consistently on Android, though presumably it could affect other platforms.

A brief investigation suggests that the order is not explicitly being set up correctly.

zea@chromium.org wrote:

    Looking at bookmark_entity_builder.cc, it seems like we're not
    explicitly setting the position when we build the bookmark
    entity. I think it's backed by a randomly generated GUID (see
    entity_builder_factory.cc), which might explain why it only
    changes on subsequent runs.

TODOs:

    * Temporarily modify the test code to be order-insensitive
    * Update the code explicitly set item position(s)
    * Update the test(s) to validate item positions

Related bugs:

    * crbug/640669
 

BookmarksTest.testDownloadMovedBookmark() expects that sync returns newly created items in LIFO order.

https://cs.chromium.org/chromium/src/chrome/android/sync_shell/javatests/src/org/chromium/chrome/browser/sync/BookmarksTest.java?rcl=0&l=163

Enabling autofill profile de-duplication, which is not otherwise expected to interact with this test, causes the ordering expectation to no longer be satisfied on re-runs of the test on a given machine.

This behaviour is only observed consistently on Android, though presumably it could affect other platforms.

A brief investigation suggests that the order is not explicitly being set up correctly.

zea@chromium.org wrote:

    Looking at bookmark_entity_builder.cc, it seems like we're not
    explicitly setting the position when we build the bookmark
    entity. I think it's backed by a randomly generated GUID (see
    entity_builder_factory.cc), which might explain why it only
    changes on subsequent runs.

TODOs:

    * Temporarily modify the test code to be order-insensitive
    * Update the code explicitly set item position(s)
    * Update the test(s) to validate item positions

Related bugs:

    * 640669
 

BookmarksTest.testDownloadMovedBookmark() expects that sync returns newly created items in LIFO order.

https://cs.chromium.org/chromium/src/chrome/android/sync_shell/javatests/src/org/chromium/chrome/browser/sync/BookmarksTest.java?rcl=0&l=163

Enabling autofill profile de-duplication, which is not otherwise expected to interact with this test, causes the ordering expectation to no longer be satisfied on re-runs of the test on a given machine.

This behaviour is only observed consistently on Android, though presumably it could affect other platforms.

A brief investigation suggests that the order is not explicitly being set up correctly.

zea@chromium.org wrote:

    Looking at bookmark_entity_builder.cc, it seems like we're not
    explicitly setting the position when we build the bookmark
    entity. I think it's backed by a randomly generated GUID (see
    entity_builder_factory.cc), which might explain why it only
    changes on subsequent runs.

TODOs:

    * Temporarily modify the test code to be order-insensitive
    * Update the code explicitly set item position(s)
    * Update the test(s) to validate item positions

Related bugs:

    * http://crbug/640669