BSD.local.dist - share/locale

[Andrew Pantyukhin]

On 8/24/06, Stanislav Sedov <ssedov at mbsd.msk.ru> wrote:
> On Fri, 25 Aug 2006 00:09:33 +0400
> "Andrew Pantyukhin" <infofarmer at FreeBSD.org> mentioned:
> > Again, you don't really mean it. Try installing any p5 port into
> > a non-localbase prefix, and you'll see that lib/perl5 and Co. will
> > be left over after deinstall. Even if you suppose that all ports
> > respect PREFIX (which many of them don't), it's not a simple
> > task to deal with non-mtree left-overs. I would very much like to
> > think that your locale effort is a step towards a better world,
> > but it really appears to be a step nowhere at all. I hope I'm too
> > short-sighted and mistaken.
>
> Didn't I say the same? ;-) You are not quite right, most of ports
> respect PREFIX and remove all directories they creates.

Okay, did you really try a full run on a current ports tree? When
you're talking thousands "most" does not mean "all but a tiny
part". There are plans to add PREFIX-related checks to pointyhat,
but until that happens, battling for clean plists is a lost cause.

> I can't really
> say why perl/ruby ports don't do this - it may be inaccurate commits, PRs
> etc. When creating ocaml framework I've added stubs to automatically
> add shared dirs into PLIST. Probably, ruby and perl frameworks require
> the same step forward.

They don't do this because they've never done it before. It's in
our history books. I'm not saying that what you're striving to do
is wrong, I'm just saying you're going the wrong way. We've
been trying to move perl stuff out of b.p.m for over three years
now, and we'll do it eventually, but you should understand
that neither thousands of problem reports, nor clear vision can
solve a problem all by themselves. It's coordination of efforts,
continuous discussion, painstaking testing and lots of other
things that make it all happen.

But I'm not trying to get personal here, Stanislav. I can only
hope your future mentor will teach you right from wrong and
direct your exceptional vigor best. I'm just trying to start a
discussion and maybe together we'll find an "open shortest
path".