Consequences of not calling CefShutdown on SIGKILL?

[snowp]

I've got an application where under certain circumstances we're forced to kill the entire process (using SIGKILL on windows), which immediately terminates the process without calling into our shutdown logic (which includes CefShutdown). Is CEF expected to handle this situation gracefully? Is there any risk of CEF leaving any persistent data (e.g. the disk cache) in a corrupted state?

[magreenblatt]

Is CEF expected to handle this situation gracefully?

No, how could it?

Is there any risk of CEF leaving any persistent data (e.g. the disk cache) in a corrupted state?

Yes.

[snowp]

Is CEF expected to handle this situation gracefully?

No, how could it?

I was thinking it might be able to detect the improper shutdown and deal with it when it was relaunched, but in retrospect that seems like it would be pretty difficult.

What kind of consequence would you see in a fresh instance of CEF using the same cache following a SIGKILL (or a crash)? Are we just dealing with potential cache misses in the new browser, or would it be best to manually wipe the cache to ensure that the potentially corrupted state affects the new instance?

[magreenblatt]

What kind of consequence would you see in a fresh instance of CEF using the same cache following a SIGKILL (or a crash)? Are we just dealing with potential cache misses in the new browser, or would it be best to manually wipe the cache to ensure that the potentially corrupted state affects the new instance?

I wouldn't expect any serious issues (e.g. crashes) or restart. You might loose cookies, localstorage, etc., if they weren't successfully persisted to disk before the process terminated -- that's something Web apps should be designed to recover from. If any cache files become corrupt the browser should be reasonably good at replacing or recovering them.