The chromium port does not currently enable page cache support. With the default process model in Chromium (Process per site instance), the page cache may not work as expected because a new process may be created when navigating to a new site instance. However, page cache would work correctly with minor modifications for other process models (process per tab or single process). Given that chromium supports different process models and that page cache would be beneficial for at least some of them. I propose to implement page cache support in chromium port.
Created attachment 194197 [details] Full WIP patch for reference Posting a full WIP patch here for reference. The idea is to split it and get review in separate sub-bugs though.
Useful manual test case: http://www.twmagic.com/misc/cache-nocache.html
Created attachment 194216 [details] Full WIP patch for reference
Created attachment 194270 [details] Full WIP patch for reference Revert WebHistoryItem changes as this should probably be solved on Chromium side, not WebKit side.
It's not clear to me that we want to support the PageCache in Chromium. We should discuss that first before landing these patches.
(In reply to comment #5) > It's not clear to me that we want to support the PageCache in Chromium. We should discuss that first before landing these patches. Adam, can you help the discussion by listing your concerns ? The combination of the existing --enable-fastback chromium command line options, the open chromium bug with a few dups and some related patches landed (e.g. r110998) let us believe that page cache for chromium would be a value at least conditionally (in case it is not enabled by default).
I'm about to go on vacation. Perhaps we should discuss the topic on chromium-dev when I get back? My main concerns are memory usage, complexity, and the interaction with the multiprocess architecture.
Shift all discussion to the Chromium bug.