Re: [Gzz] Asko 2002-11-11 (pegboard)

[Tuomas Lukka]

On Mon, Nov 11, 2002 at 07:20:00PM +0200, Asko Soukka wrote:
> 8h workday
>  - enhanced pegboard
> 
> I tried to create failing test for old non-null-string bug with Benja's
> JVM (and test did those jstring2unicharvector functions actually fix
> anything) , but I run into problem:
> 
>    Because typedef std::basic_string<unsigned short> unistring; had been
>    replace with typedef std::vector<unsigned short> unicodecharvector; I
>    had to create a new funtion using the older unistring. But std::basic
>    <unsigned short> didn't run anymore. I got LinkError from Java and
>    c++filt _ZNSt11char_traitsItE6lengthEPK (err's undefined symbol)
>    returned "unsigned short". Well, why unsigned short is not a
>    valid type anymore?

It's not that; it's that the C++ template library doesn't define a character 
traits
specialization for unsigned short.

I'm not sure what would be the best course of action.

>    I just would like to close that case before I forget all about it :)
>    Or in other words, maybe I cant forget it before I have closed it ;)
> 
> Second, about CoordSys::nonlinearity() (it was in need of development?)
> 
>    Javadoc says: "The return value is 1/l where l would be a reasonable
>    length for dicing. Returns 0 if dicing is required."
>    Should it be: "Returns 0 if dicing is _not_ required."?

Absolutely. Please fix. Oops...

>    What nonlinerity() does in TransformCoordSys-template class?
>    (I'm _not_ questioning, I just haven't figured it myself yet.)

Well, TransformCoordSys represents one transformation, on top of the
parent transformation. So the nonlinearity depends on both the 
nonlinearity of the current transform and the parent transform.

The formulas may be (and *VERY* probably are) quite incorrect.

        Tuomas