Re: email data type first release
От | Tommi Maekitalo |
---|---|
Тема | Re: email data type first release |
Дата | |
Msg-id | 200405171217.41093.t.maekitalo@epgmbh.de обсуждение исходный текст |
Ответ на | Re: email data type first release (Greg Stark <gsstark@mit.edu>) |
Ответы |
Re: email data type first release
|
Список | pgsql-hackers |
... > > Another thing is that it might make more sense to sort email addresses by > domain first (case insensitively of course), then by left hand side (case > sensitively). Since the domain is really the "most significant bit". This > is also convenient for many systems like email since they perform better > when they can handle data in that order. > > Note that this would make the type sort differently from its text > representation. This shouldn't really be a problem but occasionally you see > poorly written queries that introduce extra type conversions that the user > doesn't expect. But then if it behaves just like the text datatype then > there wouldn't be much point in using it. Sorting should then be done by top-level-domain first. Then 2nd, 3rd... and last by user. b@a.domain.com < a@b.domain.com and b@b.a.com < b@a.b.com we get then the order: a@a.a.com < a@b.a.com < a@a.b.com < a@b.b.com rather than (in normal text-order): a@a.a.com < a@a.b.com < a@b.a.com < a@a.b.com Tommi
В списке pgsql-hackers по дате отправления: