Re: localizing Pod

On Thu, Jan 30, 2003 at 03:34:29AM -0900, Sean M. Burke wrote:
> I want to make sure I understand what's being suggested here.  I've been a 
> distracted lately, so I'm merrily doublechecking.

I've also been distracted lately, hence the late reply.

Your understanding of the first proposal is perfectly correct.

> (What do people think about my adding a rule that if you have a =for 
> encoding, it has to be the first Pod directive in a file?  And the 
> corollary that you can't have two =for encoding's in a file.)

I think it'll cover 99% of the cases, so I'm fine with that.

> 2) Get perldoc (Pod::Perldoc) to add a language "infix" to the filenames it 
> searches for pod under.  So assume that currently, when I do "perldoc 
> Foo::Bar", it looks across @INC (etc) for:
> EACH-INC-DIR/Foo/Bar.pod
<...>
> Autrijus's suggestion, as I understand it, is simply to be able to say to 
> perldoc "hey, I like fr" (via a switch and/or env var), and have its search 
> be for:
> EACH-INC-DIR/Foo/Bar_fr.pod
<...>
> Right?

Correct.

> (Well, leaving aside the question of Bar_fr.pod versus Bar-fr.pod versus 
> Bar.fr.pod -- later on, we'll pick one way and stick with it.)

I'm inclining toward Perldoc.fr.pod.  If the OS doesn't support multiple
periods in the filename, it probably can't handle perlfunc_zh-tw.pod
either.

> I'm tempted to say that this second suggestion seems fine to me as long as 
> this does NOT turn into a mess where anyone expects his "I like'fr" to 
> match the Foo/Bar-fr-ca.pm, or "fr-ca" to match the Foo/Bar-fr.pm.

I think it's reasonable for "I like en" to match en-us, but not vice versa.
(Technically, I'd like to glob for DOCNAME.LANG*.pod. (And yes, I've read the
Language Detection arguments, but I'm biased this way because most
people can't process 'zh' when they request 'zh-tw', but most people can
process zh-tw *and* zh-cn if they request 'zh'. (This has to do with the
fact that Big5 is a subset of GBK (and also GB18030.))))

> I'm also tempted to say that these language tags must must must be 
> lowercase, always, under penalty of agonizing death.

Much agreed.

> And also that it if you provide a Bar-fr.pm, you must mention it in Bar.pm,
> so that no-one can say "But how was I supposed to know that file was there?
> Read the MANIFEST?!"

Change 'must' to 'really really should' and I'm fine.  It is possible (altough
not likely) that there's no Bar.pm at all.

> Does this sound reasonable?

Yup.

> Did I understand everything right?

Yes.

> And: suggestion 1 and suggestion 2 are totally independent ideas, right?

Aye.

Thanks,
/Autrijus/