Problems with unicode on select statement

[ander2]

I've defined a VARCHAR(4000 char) column which I've filled with ñ and á characters. If I set my NS_LANG=.UTF8 and run a select on the table, I'm able to retrieve 2000 characters from that column and I'm able to see the characters.

If I set NLS_LANG=SPANISH_SPAIN.WE8MSWIN1252 and run the same select, I retrieve the 4000 characters but there's no way to see the characters ok (they're a bunch on symbols).

It looks like there's some kind of WIN1252 - UTF8 conversion problem.

Could anyone help on this?

[paulish]

Have the same problem with cyrilic characters.

The field is defined as VARCHAR2(30), field data in WIN1251 codepage is 27 characters long. NLS_LANG is AMERICAN_AMERICA.UTF8. I get truncated data with select statement - 16 characters only. Doing typecasting in select statement helps (CAST(TRUNC_FIELD as VARCHAR2(255)))

I looked through the package code and found no place where field size is used for string fields. Can someone point me to this place?

[paulish]

I tried to set NLS_LENGTH_SEMANTIC to CHAR as some internet guides suggests and this does not help as well.

Not sure but perhaps similar php bug discussion can help: https://bugs.php.net/bug.php?id=54379

[paulish]

The solution is to use SetDataBuffer for string columns. Seems that occi internal buffer does not allocates enough memory for utf8 data and as result getString() calls return truncated strings. My experiments with replacing of getString() calls with SetDataBuffer() succeeded. I will cleanup the code and prepare a patch later.