Re: UTF8 is a better option than US-ASCII for the default encoding of PGStream

From: Álvaro Hernández Tortosa <aht(at)8Kdata(dot)com>
To: pgsql-jdbc(at)postgresql(dot)org, Chang Chao <charleschung(dot)cn(at)gmail(dot)com>
Subject: Re: UTF8 is a better option than US-ASCII for the default encoding of PGStream
Date: 2015-10-13 09:18:33
Message-ID: 561CCC69.8030306@8Kdata.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-jdbc


On 13/10/15 10:04, Chang Chao wrote:
> After some googling,found that the same issue was already discussed
> before.
> http://postgresql.nabble.com/JDBC-Default-Encoding-td2172427.html
> So to prevent the error message text garbling,we have to change the
> value of lc_messages from 'ja_JP.UTF-8' to "en_US.UTF8"?

Hi Chang!

Based on
http://www.postgresql.org/message-id/4678.1438350389@sss.pgh.pa.us, what
I'd say is that the best thing to do should be to send *in the startup
message*:

- client_encoding: desired client encoding
- lc_messages: C

and encode the startup message in US_ASCII (lowest common
denominator). And of course switch the encoding of further messages if
server replies (ParameterStatus) with a different encoding (which should
happen if we sent in StartupMessage a different encoding request).

Hope this helps. Cheers,

Álvaro

--
Álvaro Hernández Tortosa

-----------
8Kdata

>
> On Tue, Oct 13, 2015 at 4:48 PM, Chang Chao <charleschung(dot)cn(at)gmail(dot)com
> <mailto:charleschung(dot)cn(at)gmail(dot)com>> wrote:
>
> Hi,all
> The problem I encountered is that,when lc_messages = 'ja_JP.UTF-8'
> is set in postgresql.conf,and initial connection to database
> server failed(for example:user name,password or dbname is
> misspelled),I see garbled error message text in the client .The
> reason is that,in PGStream constrcutor,default encoding of ASCII
> is set(Encoding.getJVMEncoding("US-ASCII")).For the following
> reasons,maybe UTF8 is a better option for default encoding.
>
> 1.When initializing connection ,jdbcclient send client_encoding
> parameter as UTF-8
> see
> org.postgresql.core.v3.ConnectionFactoryImpl.openConnectionImpl(HostSpec[],
> String, String, Properties, Logger)
>
> 2. When initial connection to server completes,jdbc client only
> receive the value for client_encoding as UTF-8
> see
> org.postgresql.core.v3.ConnectionFactoryImpl.readStartupMessages(PGStream,
> ProtocolConnectionImpl, Logger)
>
> if (name.equals("client_encoding"))
> {
> if (!value.equals("UTF8"))
> throw new PSQLException(GT.tr("Protocol error. Session setup
> failed."), PSQLState.PROTOCOL_VIOLATION);
> pgStream.setEncoding(Encoding.getDatabaseEncoding("UTF8"));
> }
>
> So there seems no reason to initiate the encoding as US-ASCII.
>
> Best regards.
> Chao.
>
>

In response to

Browse pgsql-jdbc by date

  From Date Subject
Next Message Dave Cramer 2015-10-13 12:34:43 Re: UTF8 is a better option than US-ASCII for the default encoding of PGStream
Previous Message Chang Chao 2015-10-13 08:04:44 Re: UTF8 is a better option than US-ASCII for the default encoding of PGStream