Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 2 Nov 2000 17:35:43 +0800
From:      Jing-Tang Keith Jang <keith@bsdvm.jtjang.idv.tw>
To:        freebsd-ports@FreeBSD.org
Cc:        "Michael C . Wu" <keichii@iteration.net>
Subject:   Re: Something about ports/chinese
Message-ID:  <20001102173543.A5967@bsdvm.jtjang.idv.tw>
In-Reply-To: <20001102010730.B9753@peorth.iteration.net>; from keichii@iteration.net on Thu, Nov 02, 2000 at 01:07:30AM -0600
References:  <20001102144635.B5169@bsdvm.jtjang.idv.tw> <20001102010730.B9753@peorth.iteration.net>

next in thread | previous in thread | raw e-mail | index | archive | help
On 11/02/00, Michael C . Wu wrote:
> Why do we keep maintaining outta-ports when the FreeBSD base distribution
> has a Ports structure?  The whole Ports system is full of files/patch-*
> that can be construed as "ugly hacks."  Locale fixes and patches
> to make things compile do not really make that much of a difference.

The first reason that outta-port was setup two year ago was that there
were too many mini-Howtos for a Chinese FreeBSD environment.  It was
just too painful for normal users.  For example, back in that time,
if you want to fully utilize Netscape, these steps are necessary:

Install chinese/big5fonts, modify font.alias's encoding from big5.eten to big5
Build a GLGR-patched libX11 that can handle Big5 encoding correctly
Install www/libxpg4-ns, create a crt0.o for a.out building
Add entries in ~/.Xdefaults and ~/Netscape
Setup LC_CTYPE LD_PRELOAD in the environment
Use xcin25 or old xcin2.3+xa+cv to input Big5, please see another Howto :)

It just scared away many potential ordinary users.  Not everyone uses
FreeBSD only as powerful servers, it can be a desktop for daily use, just
a lot of tuning is involved for Big5 computing.  They eventually turned to
CLE(Chinese Linux Extention), which has all the mess and hacks builtin. :-)
One of outta-port's goals was to automate these mess.  In addition, many
of these has a lifetime that is so short we can even predict when it is
not needed, like X.language and xa+cv.  If we put them in ports/chinese,
it's possible that they become obsolete very quickly.  There were 2500-3000
ports in the base system, and it's unreasonable to add soon-to-be-dead ones.

The second reason was that we didn't like the way ports/{japanese,korean}
goes.  I really admire and appreciate the pioneer I18N work by these groups,
and often sees them as my best references.  But it's also true that the
port layout is mixed in the worst case.  There are no j/x11* or j/print,
but they are all put together in the same directory instead.  Although
outta-port didn't deal with this efficiently(just decreases the number in
ports/chinese) and it seems now that ports/chinese also goes in the same way
inevitably, this is a problem we eventually have to face, unless more ports
become I18N-compliant. :-)

But time is changing.  With more and more applications following I18N
standards, it's conceivable that outta-port should be integrated into
ports/chinese.  I think it's the right direction.

> IMHO, some outta-ports ports are clean enough, especially your own ports,
> Keith. :)  If it compiles and it runs, why can we not have it in the
> Ports?  ports/chinese/mutt has hacks too, and so does chinese/big5con.
> What defines a "clean" patchset?

If it compiles and runs and useless like xfig and lyx, I won't consider
them as good ports/chinese candidates. :-)  They can only accept Chinese
character input, but without good TeX or PS font backends, displaying is
the best they can do.

If it has hacks but works like CJK/big5fs/arphicttf, I think it deserves
a ports/chinese seat.  big5fs will definitely be obsolete once kernel
iconv fs interface is out, but it's quite useful for the time being.

> Would sending stuff into the FreeBSD Ports give our "hacks"
> a better chance of getting reviewed?  Certainly there are many many
> more people that can review the ports if we allow them a easy
> way of reviewing them.  Keith had to explain what outta-ports
> is just right above my reply.  Should we not let other language
> groups review our progress?

It seems like I'm an stubborn oldman who doesn't share his magic flying
blanket. :-)

I only explained the validity of outta-port in the past.  I agree with
the integration, but not just dumping all of them to ports/chinese.

> The FreeBSD system is monolithic, we do not want to fork and fork
> over and over again like Linux.  If every language locale group
> had its own "outta-ports," imagine the users and maintainers' difficulty
> of keeping up with them all.

Don't blame outta-port.  Blame those people who invented the shitty Big5
encoding, which follows no standards like ISO2022 and Unicode, and makes
Big5-ization harder in nature.

And also blame those companies who make it popular. :-)
-- 
Keep it simple and stupid.


To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-ports" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20001102173543.A5967>