On Tue, 2011-11-29 at 03:32 +0100, Lukasz Mach wrote: > (27.11.2011 22:51), Lex Berezhny wrote: > > On Sun, Nov 27, 2011 at 3:22 PM, Lukasz Mach<[email protected]> wrote: > >> It's small pyjamas snippet in existing website, and doing client-side > >> gettext is, not trivial, but not hard. And I don't think I really don't > >> want do that. It will be few strings, maybe 15? > >> > >> Doing translation on server-side, well, sounds very complicated. > >> > > If it's only 15 strings than doing anything automated seems overkill. > > I counted, it's about 30. So far, written as _("english text"), with _ = > lambda x:x - waiting for doing real translation. > > It's enough to be too hard to write such dict by hand. > > > Don't spend hours building a framework to save yourself from a few > > minutes of repetitive typing. > > Automation I imagine, is doing a bit of perl/sed on .po file, to change > it to .py dict. > > I only hoped that someone did it already (in gettext way). I18N from > examples/libtest looks like something I don't like (msg not in "human > language" like _("invoice number %s")%number, but > somemodule.label_of_invoice()%number ). > Just for the records, there's a reason for using keys (in this case methods) in stead of 'English' words/sentences. In some languages the position of the words/sentences on the screen can lead to different translations. As long as you stick to western languages, chances are not very high that you run into this issue. You can of course use different gettext domains if you use two times the same word/sentence, but that's a misuse of domains. I've done some reading on this some time ago and have lost track references for the i18n issues of gettext.
However there's a need for gettext on the list, so it would be very nice if you could get gettext working.

