I run Linux as my primary OS for most everything. I pretty much only reboot to Win2k if I need to test something. This summer I started dogfooding Thunderbird and Firebird. I really liked the fact that I could put my personal toolbar next to my menus in Firebird since it gave me more screen space for the web. There were a number of other things about Firebird and Thunderbird that really made me enjoy using them. However, the big exception (under Linux) is that link clicks from Thunderbird to http:// URLs and the like simply wouldn't work. Those who know what I'm talking about know that it sucks ass, and it pretty much makes Thunderbird extremely painful to use for day-to-day stuff. Not good. To make matters worse, for a long time drag-n-drop didn't work in the GTK2 builds either.
So, about two months ago I switched back to Seamonkey. Amazing... I started keeping on top of my bugmail because it was so easy to flip to a bug from my inbox. But, I missed some of the cool things in Firebird and Thunderbird.
Well, today I landed a patch to fix this mess. A big thanks to bz, blizzard, and bryner for reviewing the changes. Starting with tomorrow's trunk builds (and 1.6 beta), you should finally be able to click links in Thunderbird and have them load correctly in Firebird! See bug 226071 for details. Check out tomorrow's builds, and make noise if you find problems ;-)
Better late than never... bug 92928.
started doing some more IPC hacking today. looks like there is a renewed effort to enable profile sharing in mozilla. personally, i think it is a feature that makes IE very compelling to embedders. it really should be possible to embed mozilla, or to instantiate several mozilla processes, without having to worry about profile details. i'm not sure what the timeframe will be for complete, transparent profile sharing, but it seems like things are definitely starting to pick up some steam...
and so, i decided to run TestIPC today, and noticed that it has a bunch of problems. under linux, for example, the IPC daemon wouldn't recognize a client disconnect if the client was the parent of the IPC daemon! turned out this was caused by the child process (the IPC daemon) inheriting the socket used by TestIPC to talk to the IPC daemon. what a twister! anyways, the short solution was to only open the client connection to the daemon after the daemon has been spawned. as redfive is fond of saying, "good times!" :-/
anyways, i put together some much revised unix client code, which should work much better. also, bsmedberg offered to land the ipc branch on the trunk tomorrow! things are definitely on the move again :)