17.000 classes involved to migrate to Unicode????

I dont really understand yet. Does that mean:

Despite reflection and clean Smalltalk structure the CUIS (and Squeak/Pharo) - team needs reviewing 17.000 classes to bring this old Smalltalk up to date for being able to communicate with modern frameworks flawlessly?

Tnx, GS

_______________________________________________
Cuis mailing list
[hidden email]
http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org

Guido

I would suggest that you actually have a look at the code and the
discussions we had before to see what we are talking about.

Any language / system / IDE which has been around for 30 plus years
needs maintenance.

We know that you have been excluded from the Pharo mailing list
because of trolling attempts.

Regards

Hannes

On 2/7/13, Guido Stepken <[hidden email]> wrote:
> I dont really understand yet. Does that mean:
>
> Despite reflection and clean Smalltalk structure the CUIS (and
> Squeak/Pharo) - team needs reviewing 17.000 classes to bring this old
> Smalltalk up to date for being able to communicate with modern frameworks
> flawlessly?
>
> Tnx, GS
>

_______________________________________________
Cuis mailing list
[hidden email]
http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org

On Thu, 7 Feb 2013 19:57:34 +0100
Guido Stepken <[hidden email]> wrote:

> I dont really understand yet. Does that mean:
>
> Despite reflection and clean Smalltalk structure the CUIS (and
> Squeak/Pharo) - team needs reviewing 17.000 classes to bring this old
> Smalltalk up to date for being able to communicate with modern frameworks
> flawlessly?

Despite historical rough edges, Cuis is quite clean compared to Unicode, where you need language and locale information just to sort strings.  If you read through the unicode standard and related implementation information there is a huge amount of implementation complexity (e.g. http://unicode.org/reports/tr29/ describes how to determine word breaks).

You would probably find, on looking closely, that "modern" frameworks are not themselves flawless and the better ones contain a huge amount of code, e.g. to properly lay out Devengari ligatures.  If you don't support Devengari, you cut out a billion people from your target market.  A billion here, a billion there, pretty soon your market has shrunk.

As some form of English has become the international language, it is critical to do the support Cuis now has.

IMHO, a full Unicode implementation is a larger task than the full implementation of Cuis, Squeak, or Pharo.  Smalltalk makes a lot of this work much easier than many other languages, but it is still a massive amount of work.

It takes a long time to become familiar with Unicode and its issues.  E.g.
Chinese fonts are huge and full Unicode is missing a number of characters used in historical texts.  

You should start at:
        http://www.unicode.org/versions/Unicode6.2.0/

Again, this is a distraction from moving forward toward a scalable Morphic, which I consider to be of the highest value.  So we are spending time talking about what and how to do this best with limited resources and little distraction.  What level of support is required/helpful? What order of implementation? What implementation strategy? What is the least cost w.r.t. implementor time?  Do we even need this now?

It is a good planning exercise.  Planning is important.  

Anything we do should carry its weight.  Let's not do what we won't need this year.  Keep it simple.  Make it modular/add-on.  Do we need this now?  Who will sign up for the implementation work...

$0.02,
-KenD

_______________________________________________
Cuis mailing list
[hidden email]
http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org