Bug #2466 utf8 charset and size of columns

• Submitted: 20 Jan 2004 23:15
• Modified: 6 Feb 2004 10:08
• Reporter: [ name withheld ]
• Status: Can't repeat
• Category: MySQL Server
• Severity: S3 (Non-critical)
• Version: 4.1.1-alpha
• OS: Linux (Linux / Intel)
• Assigned to: Dean Ellis
• CPU Architecture: Any

[20 Jan 2004 23:15] [ name withheld ]

Description:
When using utf8 encoding, size of columns is something like size in latin1 
divided by 3 - BUT this should be rounded UP when doing mysqldump of database. 
Otherwise, you get smaller fields than you had before and possibly it won't 
hold all the data you had there. 
This is not too serious problem - you run into it only when you create 
database using different encoding and then switching encoding to utf8 later, 
but at least in mysqldump it should be fixed. 

How to repeat:
create column with size varchar(32), fill it with data, switch encoding to 
utf8, dump the table using mysqldump, import it back into mysql and compare 
the data...some data will be left out.

[6 Feb 2004 10:08] Dean Ellis

I cannot repeat this using the current 4.1.2 development sources.

It would be best if you supplied a complete test case, with the SQL statements to create, populate, and alter the table, to ensure that I am testing precisely what you are describing.

Thanks.