Re: tlug: Introduction; Looking for Help on i18n

[Kai Wetzel <k.wetzel@example.com>]

Stephen J. Turnbull wrote:
> 
> >>>>> "Kai" == Kai Wetzel <k.wetzel@example.com> writes:
> 
>     Kai> There are some things which we have very little knowledge of
>     Kai> and it would be cool if you people could help out, here:
> 
>     Kai> - Programming support for i18n within X11 (Xlib) as well as
>     Kai> C++(and C).  (Since this is what the LIP uses primarily)
> 
> All versions I have used of XIM for X11R6 are buggy.  X11R6.1 is very
> hosed on Sparc/Linux to my personal knowledge.  It is reputed to be
> very bad on i386/XFree86 as well, but this I don't know from personal
> experience.  XFree86 3.3's bugs (that I know of) can be worked around
> (although I haven't tried this with glibc yet).

The LIP has to determine what we can provide in the
library and what would be out of scope or is beeing
worked on by other people.  I can't say much about this
myself, since I only know that my Linux/X11 has
vertially no support for anything outside of the "C"
locale :(

> C/C++?  Support?  NOT.  There is some support for POSIX locales in
> Linux glibc, but no documentation I can find (I haven't looked that
> hard, and have not joined the ML, though)

Well, there is e.g. the class "locale" and there seems to
be some other facilities.  I don't know how complete they
are and I don't know how much of them is usable in common
implementations of the standard C++ library.

I am personally open for many solutions, and can only
say something about what I'd like from the API:
- Pretty easy to use
- As complete as possible
- Available within a reasonable amount of time
:O)

[...]
>     Kai> - Programming for i18n using Windows NT, Macintosh, and Java
>     Kai> (JDK 1.1).
> 
> JDK seems to do pretty well, but is not ready for prime time.

I think knowing about the bad/good points of this API, as
well as the other ones is a good way to don't repeat other
people's mistakes (and get some guide in general)

[...]
>     Kai> - Issues involved in programming for i18n in general.
> 
>     Kai> - Standards in this area, including UNICODE, etc.
> 
> Sell your car, buy the book.  Unfortunately the folks at ISO and
> Unicode want to make money from their standards, enough that they
> won't give away the standards so that there's a prayer that the
> standards will get implemented.

Sounds bad :(

>     Kai> We will need people who would like to discuss the design of
>     Kai> i18n classes and integrating them with our existing classes,
> 
> i18n _classes_?  You've already missed the boat....  If the
> functionality isn't embedded deep within the classes you already have,
> nobody is going to bother to retrofit.

Well, I think both is needed, extending existing classes as
well as potentially introducing classes which are mainly
there to provide i18n features.  Examples from the Java
API would be Locale, Collation (was: "TCollator"), Date/Time
formats (was: "formatter"), etc.  Of course some things
may effect the core classes fundamentally, such as fonts.

> Look at the Java API (java.io,
> I think) for hints; I don't like it very much (probably from lack of
> use), but you'll see that the i18n features are implemented as filter
> methods on the base classes.  Right way to go, IMHO.  Unless you plan
> to implement multiple inheritance and have the i18n classes override
> the base ASCII or Latin-1 implementation.  But that's pretty
> error-prone.

I'm pretty open here.  I have _some_ ideas how a system
which is properly internationalized should react to
the user in general (I have no clue about specific
scripts/countries, etc.)  E.g. it
should be possible to add support for, say, Cyrillic
without recompiling every application, of course.

>     Kai> as well as people who would possibly like to help implement
>     Kai> this design when it's done.
> 
> Ah ... no, I don't think so; I've got a couple of other projects on
> the fire.  I'll take a look at the web page one of these days, though.
> Maybe I'll change my mind.

That would be cool.  Even if you only help with the design, this
would help quite a lot.

kai

Next TLUG meeting is Saturday Dec. 13, 1997  (possibly Nov. 13?)
---------------------------------------------------------------
a word from the sponsor:
TWICS - Japan's First Public-Access Internet System
www.twics.com  info@example.com  Tel:03-3351-5977  Fax:03-3353-6096