Thread: is it a bug ?

is it a bug ?

From

Zouari Fourat

Date:

03 March 2005, 11:15:22

Hello,
When selecting (from php) using this select :

SELECT msgid,content,timecreated,soa FROM sms_mo WHERE (LOWER(content)
LIKE 'club%') AND (da='87000') AND (timecreated > '2005-02-18
17:00:00') ORDER BY msgid ASC

I get this error :
PANIQUE: ERRORDATA_STACK_SIZE exceeded server closed the connection
unexpectedly This probably means the server terminated abnormally
before or while processing the request.

when applying that select into phppgadmin interface there's no error.
but when commenting the line :
#client_encoding = LATIN9

in /var/lib/pgsql/data/postgresql.conf i will no longer have an
error... but got problems with charsets (i use frensh accents and arab
characters)

Re: is it a bug ?

From

Zouari Fourat

Date:

03 March 2005, 11:15:59

here's what logs contain :
ATTENTION:  Laisse de côté les caractères UTF-8 inconvertibles 0xc389
ATTENTION:  Laisse de côté les caractères UTF-8 inconvertibles 0xf474e9
ATTENTION:  Laisse de côté les caractères UTF-8 inconvertibles 0xf474e9
ATTENTION:  Laisse de côté les caractères UTF-8 inconvertibles 0xf474e9
ATTENTION:  Laisse de côté les caractères UTF-8 inconvertibles 0xf474e9
PANIQUE:  ERRORDATA_STACK_SIZE exceeded
INSTRUCTION :  SELECT "MsgId" as msgid,"Content" as
content,"TimeCreated" as tc,"SOA" as oa FROM sms_mo WHERE
((LOWER("Content") LIKE 'club%')) AND (("DA"='87000'))  AND
("TimeCreated" > '2005-02-18 17:00:00') ORDER BY "MsgId" ASC;
TRACE:  processus serveur (PID 4359) a été arrêté par le signal 6
TRACE:  Arrêt des autres processus serveur actifs
TRACE:  Tous les processus serveur se sont arrêtés, réinitialisation
TRACE:  le système de bases de données a été interrompu à 2005-02-26
16:11:25 CET
TRACE:  l'enregistrement du point de vérification est à 0/7BF31E0
TRACE:  ré-exécution de l'enregistrement à 0/7BF31E0 ; l'annulation de
l'enregistrement est à 0/0 ; arrêt TRUE
TRACE:  prochain identifiant de transaction : 70998 ; prochain OID : 535489
TRACE:  le système de bases de données n'a pas été arrêté proprement ;
restauration automatique en cours
TRACE:  la ré-exécution commence à 0/7BF321C
TRACE:  enregistrement de longueur nulle sur 0/7C05940
TRACE:  ré-exécution faite à 0/7C05918
TRACE:  le système de bases de données est prêt


On Thu, 3 Mar 2005 09:15:20 +0100, Zouari Fourat <fourat@gmail.com> wrote:
> Hello,
> When selecting (from php) using this select :
>
> SELECT msgid,content,timecreated,soa FROM sms_mo WHERE (LOWER(content)
> LIKE 'club%') AND (da='87000') AND (timecreated > '2005-02-18
> 17:00:00') ORDER BY msgid ASC
>
> I get this error :
> PANIQUE: ERRORDATA_STACK_SIZE exceeded server closed the connection
> unexpectedly This probably means the server terminated abnormally
> before or while processing the request.
>
> when applying that select into phppgadmin interface there's no error.
> but when commenting the line :
> #client_encoding = LATIN9
>
> in /var/lib/pgsql/data/postgresql.conf i will no longer have an
> error... but got problems with charsets (i use frensh accents and arab
> characters)
>

Re: is it a bug ?

From

Tom Lane

Date:

03 March 2005, 12:38:52

Zouari Fourat <fourat@gmail.com> writes:
> here's what logs contain :
> ATTENTION:  Laisse de c�t� les caract�res UTF-8 inconvertibles 0xc389
> ATTENTION:  Laisse de c�t� les caract�res UTF-8 inconvertibles 0xf474e9
> ATTENTION:  Laisse de c�t� les caract�res UTF-8 inconvertibles 0xf474e9
> ATTENTION:  Laisse de c�t� les caract�res UTF-8 inconvertibles 0xf474e9
> ATTENTION:  Laisse de c�t� les caract�res UTF-8 inconvertibles 0xf474e9
> PANIQUE:  ERRORDATA_STACK_SIZE exceeded

[ been expecting someone who knows more than me to step forward, but
... ]  What I think is happening here is that PG is expecting the
translated messages in your .po files to have the same encoding as
your database encoding, but they aren't.  Can you convert the .po
files to match your preferred encoding?

(Obviously this is something we need to improve...)

            regards, tom lane

Re: is it a bug ?

From

Peter Eisentraut

Date:

03 March 2005, 13:28:56

Tom Lane wrote:
> [ been expecting someone who knows more than me to step forward, but
> ... ]  What I think is happening here is that PG is expecting the
> translated messages in your .po files to have the same encoding as
> your database encoding, but they aren't.  Can you convert the .po
> files to match your preferred encoding?

No, this again means that you have to have a consistent database
encoding and LC_CTYPE.  The gettext library will automatically convert
between the original encoding in the file and the encoding at run time
as declared by LC_CTYPE.

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

Re: is it a bug ?

From

Zouari Fourat

Date:

03 March 2005, 13:35:40

excuse me but i dont understand what should i do ?


On Thu, 3 Mar 2005 11:28:50 +0100, Peter Eisentraut <peter_e@gmx.net> wrote:
> Tom Lane wrote:
> > [ been expecting someone who knows more than me to step forward, but
> > ... ]  What I think is happening here is that PG is expecting the
> > translated messages in your .po files to have the same encoding as
> > your database encoding, but they aren't.  Can you convert the .po
> > files to match your preferred encoding?
>
> No, this again means that you have to have a consistent database
> encoding and LC_CTYPE.  The gettext library will automatically convert
> between the original encoding in the file and the encoding at run time
> as declared by LC_CTYPE.
>
> --
> Peter Eisentraut
> http://developer.postgresql.org/~petere/
>

Re: is it a bug ?

From

Peter Eisentraut

Date:

03 March 2005, 21:28:46

Zouari Fourat wrote:
> excuse me but i dont understand what should i do ?

The locale you specified when you ran initdb needs to match the encoding
of the database.  The equivalent pairs are usually named like this:

Locale       Encoding

de_DE        LATIN1
de_DE@euro   LATIN9
de_DE.utf8   UNICODE

Pick the analogous pairs for you language environment.

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

Re: is it a bug ?

From

Zouari Fourat

Date:

04 March 2005, 11:31:44

if so, why do phppgadmin display data correctly and handle accents and
other unicode caracters ?


On Thu, 3 Mar 2005 19:28:30 +0100, Peter Eisentraut <peter_e@gmx.net> wrote:
> Zouari Fourat wrote:
> > excuse me but i dont understand what should i do ?
>
> The locale you specified when you ran initdb needs to match the encoding
> of the database.  The equivalent pairs are usually named like this:
>
> Locale       Encoding
>
> de_DE        LATIN1
> de_DE@euro   LATIN9
> de_DE.utf8   UNICODE
>
> Pick the analogous pairs for you language environment.
>
> --
> Peter Eisentraut
> http://developer.postgresql.org/~petere/
>

Re: is it a bug ?

From

Markus Bertheau

Date:

05 March 2005, 03:28:24

У Чтв, 2005-03-03 у 04:38 -0500, Tom Lane пише:
> Zouari Fourat <fourat@gmail.com> writes:
> > here's what logs contain :
> > ATTENTION:  Laisse de ct les caractres UTF-8 inconvertibles 0xc389
> > ATTENTION:  Laisse de ct les caractres UTF-8 inconvertibles 0xf474e9
> > ATTENTION:  Laisse de ct les caractres UTF-8 inconvertibles 0xf474e9
> > ATTENTION:  Laisse de ct les caractres UTF-8 inconvertibles 0xf474e9
> > ATTENTION:  Laisse de ct les caractres UTF-8 inconvertibles 0xf474e9
> > PANIQUE:  ERRORDATA_STACK_SIZE exceeded
>
> [ been expecting someone who knows more than me to step forward, but
> ... ]  What I think is happening here is that PG is expecting the
> translated messages in your .po files to have the same encoding as
> your database encoding, but they aren't.  Can you convert the .po
> files to match your preferred encoding?

Usually gettext handles this for you. And in fact, postgres.mo for
russian is encoded in KOI8-R and messages from the server are displayed
correctly UTF-8 database sessions under a UTF-8 locale.

Markus

--
Markus Bertheau <twanger@bluetwanger.de>

Attachment

signature.asc