Thread: Initdb -E LATIN1 fails when no multibyte support compiled in (current CVS)

Initdb -E LATIN1 fails when no multibyte support compiled in (current CVS)

From
Rene Pijlman
Date:
I've run into a problem with a fresh install from current CVS.
IIRC the same problem did not occur with the 7.1.2 release, so
this may be a bug introduced in the current CVS tree.

I've run configure _without_ --enable-multibyte. Then when I run
initdb with --encoding LATIN1 it fails with: 

:initdb:
:/home/rene/scratch/PostgreSQL/installed/bin/pg_encoding: 
:No such file or directory
:initdb: pg_encoding failed
:
:Perhaps you did not configure PostgreSQL for multibyte support
:or the program was not successfully installed.

Indeed when I configure with --enable-multibyte and rebuild,
initdb --encoding LATIN1 works fine.

But now the problem: LATIN1 is of course a singlebyte encoding.
So why is multibyte support needed to use it?

If this is the intended behaviour, I think the documentation of
--enable-multibyte in the INSTALL file (and perhaps in other
places) should be fixed. And if at all possible, it should be
renamed too.

Regards,
René Pijlman


Re: Initdb -E LATIN1 fails when no multibyte support compiled in (current CVS)

From
Peter Eisentraut
Date:
Rene Pijlman writes:

> But now the problem: LATIN1 is of course a singlebyte encoding.
> So why is multibyte support needed to use it?

--enable-multibyte really gives you two things:

1. ability to handle multi-byte characters sets in string mashing
functions

2. character set conversion between client and server

These things are technically unrelated but the group of users that need
this seems to have coincided.  If you want to propose splitting this up it
could be discussed.  Maybe #2 could even be on by default.

> If this is the intended behaviour, I think the documentation of
> --enable-multibyte in the INSTALL file (and perhaps in other
> places) should be fixed. And if at all possible, it should be
> renamed too.

I think the INSTALL file refers you to the Admin Guide for more
information, where this is explained.  Maybe that needs to be improved.

-- 
Peter Eisentraut   peter_e@gmx.net   http://funkturm.homeip.net/~peter



Re: Initdb -E LATIN1 fails when no multibyte support compiled in (current CVS)

From
Rene Pijlman
Date:
On Sun, 12 Aug 2001 12:50:57 +0200 (CEST), you wrote:
>1. ability to handle multi-byte characters sets in string mashing
>functions
>
>2. character set conversion between client and server
>
>These things are technically unrelated but the group of users that need
>this seems to have coincided.  If you want to propose splitting this up it
>could be discussed.

I have no reason to.

>> If this is the intended behaviour, I think the documentation of
>> --enable-multibyte in the INSTALL file (and perhaps in other
>> places) should be fixed. And if at all possible, it should be
>> renamed too.
>
>I think the INSTALL file refers you to the Admin Guide for more
>information, where this is explained.  Maybe that needs to be improved.

The INSTALL file says:
"--enable-multibyte
Allows the use of multibyte character encodings. This is
primarily for languages like Japanese, Korean, and Chinese. Read
the Administrator's Guide for details."

This is misguiding for users with singlebyte (e.g. European)
encodings. Even if the Admin Guide explains it better, I think
the explanation in the INSTALL file should be corrected.

Regards,
René Pijlman