On Jul 21, 2011, at 6:16 PM, David E. Wheeler wrote: > On Jul 21, 2011, at 4:01 PM, Greg Sabino Mullane wrote: > >>> I think that setting pg_encoding should always turn pg_unicode *on*. >> >> Hm...no, I think it should always be off. If someone really wants a >> different >> encoding, they probably are used to it coming back "raw". David C, >> I think we talked about this? > > I disagree. It's me telling DBD::Pg what encoding the database uses, but I > definitely want that converted to Perl's internal form. I *only* want raw if > I explicitly ask for raw (or if there's no choice, such as when I set the > encoding to ":raw" or something).
As a developer, why would you care if this information is available from the database itself? If you are caring about the encoding at all, you would be dealing with bytes/octets. Perl does not store unicode characters in any format besides UTF-8 so you're not changing "internal" characteristics ; what DBD::Pg uses to talk to your database shouldn't matter. > I think of it being kind of like the `encdoding` pragma, in which I declare > the encoding of my source code. Perl sees that and converts it to its > internal form. The only time this would be useful would be if your database is set to something inscrutable (aka SQL_ASCII); if your end result is meant to be internal perl, you have no business providing an encoding. >>> Maybe. I think a lot of existing installations may find they need to >>> turn it off, unless they had been using pg_enable_utf8 before. >> >> Yep: no way to know until we release. David and I were thinking that the >> other direction (data going to database) is probably more likely to >> break things. > > I wonder if, as an interrime measure, existing code that sets pg_enable_utf8 > should still do something, like set pg_encoding to "utf-8" and turn > pg_unicode on. Yeah, I'd had that thought, along with spitting the deprecation warning. >>> Oh I agree, I just think it's worth putting off until this other stuff >>> gets sorted out. >> >> Nah, the more stuff we can fix out of the gate the better. > > Okay. +1. >>> Have you asked Tim Bunce about any of this stuff? I know he has >>> thought about adding encoding knobs to the DBI core, but I don't >>> know how far a long he got in thinking about a design. >> >> Good idea: I have not. Will try to do so. Or anyone else that wants to >> raise this on dbi-dev.... > > Yes, a must, IMHO. More cooks! ;-P Yeah, it'd be nice to know what at least some proposed interfaces/APIs are so we don't need to support a whole other place setting for years to come. Regards, David -- David Christensen End Point Corporation da...@endpoint.com
