Hi manphiz, Thanks for the quick reply。
On Sun, Aug 12, 2007 at 10:27:47AM +0800, manphiz wrote: > Ming Hua wrote: >> I've been using scim 1.4.7-1, with /FrontEnd/X11/Dynamic as false, XIM >> mode, and over-the-spot style for quite some time. In the past few >> weeks, I've only encountered one keyboard input lock-up, and I'm not >> even sure it's scim's fault. >> >> So I got curious (since there are many Ubuntu bug reports saying the >> lock-up is quite frequent) and started playing with my scim settings. I >> finally found out that such lock-up seems to happen much often with >> on-the-spot style. This seems also related to the "Nautilus breaks on >> several places and can't input anything" bugs (for example [1]), as I've >> never had problem before with nautilus, but seem to be able to reproduce >> them with on-the-spot style. >> >> So manphiz, have you been experiencing this bug with 1.4.7-1? And what >> input style are you using? You can check in the scim-setup window, by >> "FrontEnd -> Global Setup -> Embed Preedit String into client window" >> item. Can you test the other style to confirm my conclusion? >> >> 1. https://bugs.launchpad.net/ubuntu/+source/scim/+bug/90043 >> > I tried with /FrontEnd/X11/Dynamic set to false, and tweaking the "Embed > Preedit String into client window" in scim 1.4.7-1. The lock up did > disappeared for a lot of normal usage said to trigger the problem, but > still showed up in some of them, one of which is the "massive new tabs in > gnome-terminal", which still resulted in lockup and changing the spot > setting didn't help but can be resolved by setting /FrontEnd/X11/Dynamic to > true. The nautilus one mentioned in your link to launchpad is still > reproducible at my place too, which really compromises daily use. There may > be other way to trigger the lock up as well. After some more testing, I realized that the "Embed Preedit String into client window" thing in scim-setup doesn't seem to change the scim setting properly. After you unset this option, i.e., using "over the spot" style, what is the result of the command "grep OnTheSpot ~/.scim/config" on your system? If you get /FrontEnd/OnTheSpot as false but /FrontEnd/X11/OnTheSpot as true, can you manually change /FrontEnd/X11/OnTheSpot to false (make sure scim isn't running when you edit ~/.scim/config file) and do some more testing? Changing both to false seems to fix the lock-up problem for me in XIM mode with /FronEnd/X11/Dynamic = false. > I agree that maybe scim is not the only culprit, I'll keep testing with > /FronEnd/X11/Dynamic set to false. Based on the current information, maybe > it is reasonable to reassign this bug to libX11. Yes, I do suspect this is a libX11 bug (since Redhat people say so), but as there aren't many people familiar with both input methods and X (evident by the fact the X.org upstream bug has been silent for quite a while), I want to find out the exact procedure to reproduce this bug before reassign/clone this bug to libX11. Thanks again, Ming 2007.08.12
signature.asc
Description: Digital signature
