Amit Langote
amitl****@gmail*****
2013年 10月 25日 (金) 13:57:21 JST
On Thu, Oct 24, 2013 at 9:16 PM, Fujii Masao <masao****@gmail*****> wrote: > On Thu, Oct 24, 2013 at 3:43 PM, Amit Langote <amitl****@gmail*****> wrote: >> Hi, >> >> Is the requirement of C locale necessary? >> If I use say a European locale like sv_SE.iso88591, what behavior of >> pg_bigm could be suspicious? >> >> Is this because, most of the locale dependent ctype functions behave >> predictably under C locale and we happen to use them a lot? >> >> But, I also see that we use "tsearch/ts_locale.h". Doesn't that help >> ease that restriction? >> >> What am I missing? > > TBH, I've never encountered any problem caused by the locale > other than C. My concern is that pg_bigm uses bigmtextcmp() > when comparing two strings, but this function doesn't take into > consideration the locale. OTOH, PostgreSQL basically uses > bttextcmp() to compare two strings, and it seems to take into > consideration the locale. So, I'm not confident that no problem > would occur under the locale other than C. Thought? > What if C locale does not work for some users which happens sometimes, for example, European characters)? I am also trying to understand implications of using C locale with multi-byte encodings. Have you so far observed any incompatibilities or locale dependent misbehavior when using C for UTF-8 or EUC_JP or SHIFTJIS? -- Amit
Fork