Re: More collation questions

[Mark D. <mar...@jt...>]

For problem 1.
Omitting *intermediate* levels will not work in general. If there is any
tailoring, the meaning of each level is only well-defined if the higher
level is around. I think I supplied examples previously on this list, but I
don't have time to dig them up now.

One could modify the code to have the Identical level even if other levels
were missing. That would distinguish half-width and full-width forms, at
least by codepoint order.

For problem 2.
I wish as well that the tertiary values in the UCA were consistent regarding
case. If all uppercase letters were after all other letters, it would make
it much easier to deal with. Unfortunately, that was not done; the case
changes are interleaved among others. One could modify the code to have a
special option to mask out case differences in the UCA and tailorings. That
would take a bit more work than the stand-alone identical level above.

For both of these, if we were to look at doing any changes ourselves we'd
have to see some evidence as to why and how these features were important.

Mark
__________________________________
http://www.macchiato.com
►  “Eppur si muove” ◄

----- Original Message -----
From: "Vladimir Weinstein" <we...@jt...>
To: "Avery Chan" <av...@ap...>
Cc: "icu list" <ic...@os...>; <mar...@jt...>
Sent: Thursday, October 17, 2002 08:00
Subject: Re: More collation questions


> Hi,
>
> Both of your problems are good candidates for an RFE.
>
> In the first problem, you want to have primary and tertiary levels on,
> but secondary off. As far as I can see, you would like to pick and
> choose the strength levels for your collator. This is easy to
> accomplish, but I'm not sure it is in accordance with the UCA.
> Mark, can you advise?
>
> On second problem you agree with my fundamental grudge with the UCA:
> case differences are not distinguishable from other tertiary
> differences. Something like this is doable, although it would require a
> thorough rehaul of tertiary weigths generation. It is an interesting
> usage model - would you care to provide the motivation for it?
>
> Let me know if you have any other questions.
>
> Regards,
> v.
>
> Avery Chan wrote:
> > Last collation question for a little while:
> >
> > I made a matrix of options that I wanted. I found several places where
> > some of the collation options seemed to be in conflict.
> >
> > 1)If I want my collator to be sensitive to halfwidth and fullwidth forms
> > I use tertiary level strength, but if I want the collator to be
> > diacritic insensitive that would mean I'd need to use primary strength.
> > How do I get around this?
> >
> > 2) Similar situation but this time I want my collator to be
> > case-INSENSITIVE but be sensitive to halfwidth/fullwidth forms. Is there
> > a way to factor out the case-level sensitivity at the tertiary level?
> >
> > Thanks for all the help!
> >
> > -Avery
> >
> > _______________________________________________
> > icu mailing list
> > ic...@os...
> > http://oss.software.ibm.com/developerworks/oss/mailman/listinfo/icu
> >
>
> --
> Vladimir Weinstein, IBM GCoC-Unicode/ICU  San Jose, CA we...@jt...
>
> _______________________________________________
> icu mailing list
> ic...@os...
> http://oss.software.ibm.com/developerworks/oss/mailman/listinfo/icu
>