Re: DBD::PostgreSQL
| От | David Wheeler |
|---|---|
| Тема | Re: DBD::PostgreSQL |
| Дата | |
| Msg-id | 1A15B4B0-FB20-11D6-93B3-0003931A964A@wheeler.net обсуждение исходный текст |
| Ответ на | DBD::PostgreSQL (David Wheeler <david@wheeler.net>) |
| Список | pgsql-interfaces |
On Monday, November 18, 2002, at 09:44 AM, Tim Bunce wrote: > Just to be sure this is clear, begin_work is *just* an alternative > way to set AutoCommit off till the next commit or rollback. The > application is free to just set AutoCommit explicitly as needed. > > The *only* time a driver needs to consider the BegunWork attribute > is immediately after a commit or rollback. Okay, thanks for the clarification, Tim. > Should need to "parse" in the formal sense, no full grammar is > needed, it's just a very quick skim through the string. If done in > C it should be too cheap to worry about. Especially as it's only > done at prepare() time, not execute(). Okay. I'm used to thinking about things in Perl time, so I'm not quite sure about the expense of the c code stuff. Thanks for the tip. > Would binding a string type to an integer (etc) placeholder work? > If so, just ignore the types and bind everything as strings. > That's exactly what DBD::Oracle does. Well, it wouldn't work for bytea columns, but that's true with non-prepared statements already, anyway. Interesting idea, and I think you may well be right. Tom, can you confirm? Thanks, David -- David Wheeler AIM: dwTheory david@wheeler.net ICQ: 15726394 http://david.wheeler.net/ Yahoo!: dew7e Jabber: Theory@jabber.org
В списке pgsql-interfaces по дате отправления: