utf8网站和latin1数据库表字段

Question

I am using mysql database, innoDB and MyISAM both engines. I want to see the different between utf8 and latin1, so I did a test :
codes On website :

<meta http-equiv="Content-Type" content="text/html; charset=utf-8" />

attribute of field on database table :

latin1_swedish_ci

Then I typed chinese words "我爱你" and "ĀāĂ㥹ĆćĈ" on the website and click submit button, php and mysql store the words into database. And mysql retrive value from database and display onto the website back.

Output :
website show "我爱你" and "ĀāĂ㥹ĆćĈ".
databse table field show "我爱ä½" and "Ā".

I did test all the following changes :
1) change meta of website to latin-1 and attribute of table field to utf-8
2) change meta of website to utf-8 and attribute of table field to utf-8 too.
3) change meta of website to latin-1 and attribute of table field to latin-1 too.
but the output is still the same, the output change nothing. Why?

Is that I can't test by that method? If so, how do I test the different between utf-8 and latin-1?
How to make database table field show the words "我爱你" and "ĀāĂ㥹ĆćĈ"?

I am developed a social networking website like Facebook.com which support multiple languages, should I use utf-8 for database field? disadvantage of utf-8 is it takes 3 bytes per character but latin-1 takes 1 byte only. To save the storage, it is better to use latin-1, but I am not sure what will be the future problem if I use latin-1 instead of utf-8. Can anyone give me some advice of how to decide which character set to be used?

Answer 1

1) You may want to note that you can't really peak into a database record without using some kind of software, which may bring to the show its own bag of issues. Phpmyadmin has a character set config option, software products have an internally assumed character set, even a command prompt window has a codepage. The important thing is to make sure you get back from the database exactly what you've put in it, not how is it stored in the tablespace. Use "SET NAMES character-set" to keep consistent charset thru the whole connection.

2) UTF-8 is clearly where the world is moving to, because it works and because it can store characters from every language (writing system) you're likely to encounter. With latin-1 you are cutting out every language which is not from Western Europe - this means not only chinese and cyrillic and greek/hebrew and such, but also East Europe, Turkey and a lot of other places which basically use latin alphabets with some particular letter added on.

3) UFT-8 is, and is expected to be in future, by large the most future-proof solution.

4) It's much safer (and saner), ever for monolingual applications, to do the right thing from the start (that would be UTF-8), rather than have to convert your multi-gygabyte tables later on, when you find out you need more. Nobody who had to do such a thing liked the experience.

5) Disk space is a commodity, cheaper by the day - if you are about to do 'social' you should just take a boatload of it ( if the thing flies you'll need it anyway) and forget about it, there are other issues that will bite you much sooner than disk: performance under load, access concurrency, clustering and load balancing multiple server. I can't remember a single social network lamenting issues because of those 3 bytes.