Comment # 7 on bug 1069162 from Zhao Qiang ������

Current status is:
C��dric's system didn't produce any core-dump when libpinyin crashed. 
So I tried to artificiality make a dump point inside libpinyin, and I didn't
modify anything on tumbleweed's env, But I found it dumped successfully, which
can be found obviously by coredumpctl.
So, I guess that maybe C��dric's program didn't really dump, or maybe just exit
because of error env edition. 
If this thought is true, then the trace log will also not direct to real point
which makes the ibus-pinyin exit.

On the other side, based on the backtrace list: 
It's obvious that "m_user_bigram->store" failed, there are 2 suspect point in:
https://github.com/libpinyin/libpinyin/blob/27ba2294e985622ecfbbb5cd8eb1695b3298a388/src/storage/ngram_kyotodb.cpp#L134-L142
1), For some reason, "Bigram::attach" didn't run or failed inside. so
"Bigram::store" can not get an initialized m_db in Line:136.
2), when it copies strings to the DB memory failed, maybe strings length is too
long or the memory position point to an invalid place as Line:141.

Anyway,  For further investigation, only if I can get the dump file or the
image if he runs in a virtual machine. Because all the thing works fine from my
side, I can't get any thread by static code analysis by now.