commit | 4c31a98af9891823d226c219d034a8422d17b2cc | [log] [tgz] |
---|---|---|
author | Vardhan Mudunuru <vardhan@google.com> | Wed Jan 23 17:06:20 2019 -0800 |
committer | CQ bot account: commit-bot@chromium.org <commit-bot@chromium.org> | Thu Jan 24 05:47:33 2019 +0000 |
tree | 38a227d45fd7cda31ee768fd14e87378527f639a | |
parent | 3202e197bf4477772a4391d91fbe72208b194ded [diff] |
[modular] Fix & renable flaky tests embed_shell and story_shell There is a race condition between the story shell getting a ViewOwner for the module, and the module dropping the ViewOwner on the floor (because its a test module which doesn't care to draw things on the screen). The issue happens when the latter happens first, and the ViewOwner that the story shell receives is invalid, which causes the FIDL call to fail (since the FIDL interface requires it to be valid). Fixes in this patch: * Hold onto ViewProvider in these test modules indefinitely so ViewOwner is not dropped on the floor * unrelated: fix FindAnchor() to use ModuleData::is_embedded, since the other signal is not accurate. Additional cleanup: * Clearer names which would've helped me follow code quicker (e.g., ProcessPendingStoryShellViews) * Fix code to follow fuchsia/google C++ style guide (e.g., data types declared in the class first). MF-178 MF-174 #done TEST=embed_shell and story_shell don't flake for me anymore. Change-Id: I42250accaf301c10ec66dbdb2df37515f55408a5
Peridot is a framework for composed, intelligent and distributed user experiences.
Applications not explicitly designed to interoperate (and possibly implemented in different programming languages) are ephemerally downloaded and dynamically composed to run in a shared context. The framework manages application lifecycle, resources and view hierarchy; and well as context and suggestion infrastructure.
State of Peridot experiences is transparently synchronized across user devices using a distributed offline-first storage system.