Re: Mixing different LC_COLLATE and database encodings

Bill Moseley wrote:
> Do I have these statements correct?

yes

> 1) What else is the database's encoding used for besides to determine
> how to convert text in input and output based on the client encoding?

nothing

> 2) What client encoding is used if the client does not specify one?

the server encoding

> 3) The vast majority of my utf-8 encoded text that I need to display
> sorted probably maps to 8859-1 characters.

probably not :)

> That is, if I have text that's in utf-8 but includes characters that
> would map to 8859-1 (say accented chars), that sorting will not be
> correct because it's not converted to 8859-1 when sorting?

right

> 4) If the above is true, then if I wanted my utf-8 encoded text to be
> sorted correctly then I'd need to re-initdb using
> --encoding=en_US.UTF-8, correct?

right

> 5) I suppose there's not way to answer this, short of running
> benchmarks, but any ideas what using a lc_collate with utf-8 would do
> to performance? Is it a big hit?

I don't know why that would be a problem.

> Not related to Postgresql, but testing some of this is confusing
> due to my environment. How do I get my xterm to work with utf8?
> Does ssh do something with encoding?

I don't use xterm so I'll skip the rest.

--
Peter Eisentraut
http://developer.postgresql.org/~petere/