DBI peeps,

Sorry for the delayed response, I've been busy, looking to reply to this thread 
now.

On Sep 9, 2011, at 8:06 PM, Greg Sabino Mullane wrote:

> One thing I see bandied about a lot is that Perl 5.14 is highly preferred. 
> However, it's not clear exactly what the gains are and how bad 5.12 is 
> compared to 5.14, how bad 5.10 is, how bad 5.8 is, etc. Right now 5.8 is 
> the required minimum for DBI: should we consider bumping this? I know TC 
> would be horrified to see us attempting to talk about Unicode support 
> with a 5.8.1 requirement, but how much of that will affect database 
> drivers? I have no idea myself.

I think I'd just follow TC's recommendations here. DBI should stay compatible 
as far back as is reasonable without unduly affecting further development and 
improvement (not that there's much of that right now). So if proper encoding is 
important to you, use at least 5.12 and prefer 5.14. And if proper encoding is 
not important to you, well, it is, you just don't know it yet.

> Another aspect to think about that came up during some offline DBD::Pg 
> talks was the need to support legacy scripts and legacy data. While the 
> *correct* thing is to blaze forward and use Do Things Correctly everywhere, 
> I think we at least need some prominent knobs so that we can maintain 
> backwards compatiblity for existing scripts that expect a bunch of 
> Latin1, or need the data to come back in the current, undecoded, 
> un-utf8-flagged way.

Agreed. I suspect the existing behavior should remain the default, with a knob 
to make it "do things correctly," with perhaps a deprecation plan to turn on 
the "correctly" knob by default in a year or so.

Best,

David

Reply via email to