>>>>>>>>>>>> Monty Taylor wrote (2008-09-30 15:14:30): > So rather than build our own system of dealing with all of this - I'd > love to see us be able to use some of what's already there. Better than > the C version of this, the C++ one seems to understand you might want to > use more than just one global locale. Now, I'm not sure how charsets > enter in to this setup... but the ability is there to deal with > collations, numbers, currency and dates. Any thoughts?
To be honest, I think a database should deal with as little as possible with these things. Currency is just a number and the the fact that the number is a currency should be dealt with by the application. The same goes for dates which should be ISO 8601 compliant in the db and how to display the date in some locale should be dealt with by the application and so on (I even think that timzeones should be ripped out of all databases and dealt with by applications. Timezone handling in databases causes nothing but complexity and confusion. A good database design would be independent of locales and timezones (and thus not SQL compliant....)) When it comes to character set encoding, I think there are three important aspects to consider. 1) The on-disk encoding. Chinese and Japanese text will require less space in UTF-16 than in UTF-8. For western langauges it's the other way round. But this aspect is only relevant when the amount of text stored is large relative to the available disk space. 2) The database native (or in-memory) encoding. I think space is not as critical here. If one e.g. choses ICU to handle collation, it would be natural to chose UTF-16. 3) The interface encoding. This will be dependent on the interface chosen. E.g. If the interface is JDBC the strings are by definition in UTF-16. I think 2) should be chosen with respect to speed and the ability to get clean code. For the moment UTF-16 seems to me to be a good choice. 3) is as mentioned defined by the interface used, and if 2) is UTF-16, then 1) should by default also be UTF-16 but with the option to configure or plug-in other encodings. -- Bernt Marius Johnsen, Staff Engineer Database Technology Group, Sun Microsystems, Trondheim, Norway
signature.asc
Description: Digital signature
_______________________________________________ Mailing list: https://launchpad.net/~drizzle-discuss Post to : [email protected] Unsubscribe : https://launchpad.net/~drizzle-discuss More help : https://help.launchpad.net/ListHelp
