Hi,

I am using your japanese input method for some time now but the crash rate (when choosing other than the first suggestion) is pretty bad on my end so I am trying to track the bug down.

Using gdb, I found that it crashes in libanthy when swapping the suggestions around to make the one with the highest score in the first place and it seems the memory is in pretty bad shape within the suggestions' dictionary.

I'd like to trying a workaround for now (just commenting out the swap process) but I have a hard time figuring out how you managed to compile libanthy using the source available in your repository on github…

Could you help me?