- Digging out from under needinfos and reviews after pto from last week/3 day weekend.
- bug 1160142 (take browser stacks sooner for plugin hangs) - waiting on review still
- bug 1156857 (M6, histogram child data missing) - fixed and closed.
- bug 1137944 (flash paint bug on Windows)
- making progress on this using an approach I started experimenting with a while back - telling windows not to clip the parent window where child windows are located, and doing our own clipping to compensate.
jimicy (jimmy wang)
bug 866413, 1040947, 1162718 all related to pageInfo being sluggish on e10s
- using unsafe CPOW operations get the image nodes for the media tab on page info
- refactoring code that uses the contentWindowAsCPOW to be in a frame script
- some parts of pageInfo has arrays no one pushes to in the code and there's loop through them (could be for add-ons).
- Add-ons might break because of moving to a frameScript.
- Bug 1167594 - Make it possible for the preloaded browser to be remote was r+
- Bug 1145303 - Remove unused "What is this page?" link and related intro messaging was r+
- Bug 1167601 - Converting newTab.xul to newTab.xhtml to be rendered in content process
- more specifically migrating xul panels into the mainPopupSet (search panel working in Bug 1167643 but needs more work with search engine panel - customization panel is next in Bug 1168589)
- using remote page manager to set up messaging infrastructure between parent and content process
* 1132518 - Document navigation with F6. Existing code iterates over docshells, so have mostly rewritten this code to follow the same path as tab navigation. * 1149745 - Make <select> select options as cursor is pressed on Windows
- missed most of past week for surgery and holiday
- catching up on regular work and needInfo's
- 90% done with next iteration of arewee10syet. Which will include fourth state of "Shimmed"
- have one bit logic to sort out to make sure untested add-ons don't get shown as status shimmed. Will have it solved and pushed to production by tomorrow.
- Found the underlying issue regarding the events bleeding through from popups. Fix landed earlier this week.
- Mostly been reading up about Mach's virtual memory system and had a decent chat with markus and jeff on monday and we think we have a pretty decent plan of attack. I'm still in ongoing chats with them to work out implementation details once I have a better understanding of how vm works in Mach and how it's used in Gecko.
- Finally have review on <datalist>! Landing today.
- Attached a patch for bug 1079122 (popups from bookmarklets)
- Investigated needinfos.
- bug 1164640 - reset the a11y flag - landed
- bug 903016 - contentAreaClick - work in progress
- bug 1129957 - RemoteWebNavigation doesn't accept post headers - investigating test problems
- Bug 1164014 - Shim optimization, r+ about to be landed.
- Bug 1164011 - interpostion for CPOWS, landed
- Bug 1110123 - nsIAboutModule seems like doing it in C++
would be quite hard... thinking about doing a special shim like mechanism for this in XPCOMUtils.js
- Bug 1153256 was not fixed after all, tried something much more drastic in 22.214.171.124rc1 (currently in the Beta channel) by using nsIHttpChannel.redirectTo(). Found a couple new bugs in this approach (see below)
- Worked around redirectTo() breaking Flash plugin subrequests unconditionally (must file a bug, probably affects HTTPS Everywhere too)
- Using redirectTo(), then suspend() to do async work and resume() at end doesn't work so bad, but still a connection is made (no HTTP traffic fortunately, though). Preventing that TCP traffic would be desirable, about to file a bug
- Created a github repo for billm ( https://github.com/hackademix/noscript ), but still empty actually because directory structure and build script badly need to be overhauled before publishing (currently depend on Ant)