[#CORE-2785] BLOB transliteration problem in a specific case

Firebird Core

BLOB transliteration problem in a specific case

Created: 02/Dec/09 05:10 AM Updated: 03/Dec/09 07:59 AM

Component/s:

Charsets/Collation

Affects Version/s:

2.5 Beta 2, 2.1.3, 2.5 RC1

Fix Version/s:

2.5 RC2, 2.1.4, 3.0 Alpha 1

Time Tracking:

Not Specified

File Attachments:

2785.sql (0.5 kB)
2.

db.sql (2 kB)

Environment:

Windows, IBExpert, Client CharSet UTF8

Issue Links:

Relate

This issue relate to:

~~CORE-2222~~ AV in engine when storing text blob with transliteration blob filter

Planning Status:

Unspecified

Description

« Hide

Transliteration error occurs with statement

COMMENT ON table T IS
'параметры: количество
             если больше 0 подгоняет сумму товаров под данное значение пропорционально себестоимости
               если меньше 0, на выходе в GSUMMA будет себестоимость за 1 ед.
               если равно 0, делает только разложение на компоненты
  Результат';

--

This operation is not defined for system tables.
unsuccessful metadata update.
Create metadata BLOB failed.
Cannot transliterate character between character sets.

--

Reproduces with IBExpert. With ISQL - or hangs up, or gets error too:

Statement failed, SQLCODE = -607
unsuccessful metadata update
-Create metadata BLOB failed
-Cannot transliterate character between character sets

With CharSet WIN1251 works fine.

All

Comments

Work Log

Change History

Version Control

Subversion Commits

Sort Order:

[ Permalink | « Hide ]

Adriano dos Santos Fernandes added a comment - 02/Dec/09 09:03 AM

Cannot reproduce. Feel free to return with reproducible test case.

I guess you're using incorrect client charset, not matching the text encoding.

[ Permalink | « Hide ]

GR added a comment - 02/Dec/09 10:32 AM

>> Cannot reproduce. Feel free to return with reproducible test case.

Try to reproduce with attached file (scipt).

>> I guess you're using incorrect client charset, not matching the text encoding.

Everything is possible. But removing 1 space char query works correctly.

[ Permalink | « Hide ]

Dmitry Yemanov added a comment - 02/Dec/09 10:33 AM

I've attached a complete failing script (it's saved into UTF8 by the notepad). I can reproduce the issue using isql -i <db.sql>

[ Permalink | « Hide ]

Adriano dos Santos Fernandes added a comment - 02/Dec/09 11:07 AM

This is not direct related to COMMENT ON. It's a blob transliteration problem in a specific case.

[ Permalink | « Hide ]

Adriano dos Santos Fernandes added a comment - 03/Dec/09 07:59 AM

To make issue better documented, that is what happens.

When blob are transliterating, we consider that segments could have truncated text, in this case the truncated part is processed when the next segment is written.

But also, a transliteration may increase the string length, and if it is greated than 64K, it's written in chunks.

When the not-first chunk has truncated string, it has not being deferred to next segment and error incorrectly happens immediatelly.