Postgresql bugs (v 1.5 CVS)

[moj]

After an update (step 5) I got the following:

OPTIMIZE TABLE faqadminlog, faqadminsessions, faqcategories, faqcategoryrelations, faqchanges, faqcomments, faqdata, faqfragen, faqnews, faqsessions, faquser, faqvisits, faqvoting

I guess postgresql will need something like "VACUUM ANALYZE ..." instead.

After filling out the for for "Ask a question" I only get:

Required fields are your name, your email address and your question.

And the question is not saved.

/moj

[Thorsten]

Hi,

I fixed both bugs in CVS und beta3.

bye
Thorsten

[moj]

Thanks, seems to work ok now

Just another minor postgres bug in the generated backup sql script. The comments must begin with "--" and not "#".

[Thorsten]

Hi,

you're right. I'll fix it as soon as possible. Thanks for testing!

bye
Thorsten

[moj]

Unicode in postgres

Don't know if I have some misconfiguration, but I get problems with Swedish characters in my unicode (utf8) database.

When I try to add a category with e.g. "ä" I get the following message:
ERROR: invalid byte sequence for encoding "UNICODE": 0xe46b74

I think that the problem is that the web page with the form is ecoded as "iso-8859-1".

Is it something I have missed in the configuration or is it a missing configuration option?

/moj

[Thorsten]

Hi,

it's a problem in the Swedish language file. It has to be rewritten as utf-8 because I use the encoding of each language for the page encoding.

bye
Thorsten

[moj]

Ok, I see!

So, an additional variation with an alternate Swedish language encoded as utf-8 might solve my problem?

But isn't this a problem if I need to support several languages with different encodings. At least Postgresql will not accept to store characters outside the current encoding in the database.

Maybe a solution could be to disable the language choice on the start page. Or connect to different databases depending on language (or really language encoding), e.g. one db per encoding.

/moj

[Thorsten]

Hi,

what about a function which encodes all international characters into HTML entities?

bye
Thorsten

[moj]

Good suggestion!

But it could be som problems with comparisons and free text search. I don't know exactly how for example tsearch2 in postgres will handle & and ;, probably as non-word characters. Maybe similar problems in mysql also.

Another solution migth be to have a funtion to translate between language encoding and db encoding. Lanaguage encoding have to be a subset of db encoding ofcourse.

/moj

[Thorsten]

Hi,

I didn't implemented tsearch2 yet but it's prepared.

I will think about a solution for this problem.

bye
Thorsten