On 9/23/07, Malcolm Tredinnick <[EMAIL PROTECTED]> wrote:
>
> Previously, I've wondered about pulling string_concat out of trans_real
> and into __init__, since it doesn't depend on the USE_I18N setting at
> all. That piece of tidying would avoid some of these problems, too, so I
> think I'll do
On Sat, 2007-09-22 at 04:31 -0700, Andrew Durdin wrote:
[...]
> I'm confident now that the patch is correct, as I now understand why
> it works:
>
> string_concat returns a __proxy__ that will call _string_concat when
> __unicode__ is called, so we expect _string_concat to return a unicode
>
On Sep 22, 2:16 am, Malcolm Tredinnick <[EMAIL PROTECTED]>
wrote:
>
> Thanks, Andrew. That patch looks "intuitively" correct; it's in the
> right area of code that I would have expected to find the problem, etc.
> I'll have a think about it through today to see if there are any
> technical
On Fri, 2007-09-21 at 22:43 +, Andrew Durdin wrote:
> I was looking into #4796, and submitted a patch that fixes the error
> -- and I'm fairly sure that it fixes it in the appropriate spot.
> However, I didn't quite grok the runtime interaction between the
> __proxy__ objects and the
I was looking into #4796, and submitted a patch that fixes the error
-- and I'm fairly sure that it fixes it in the appropriate spot.
However, I didn't quite grok the runtime interaction between the
__proxy__ objects and the delayed_loader function, so I'd appreciate
it if someone else could cast