RE: [LINUX5250] Screen bleeding

[James Rich <james@xxxxxxxxxxx>]

On Mon, 7 Feb 2005, Ruel, Marc-Antoine wrote:

> I have one warning, please don't put the Tn5250Win[] and following 
> variables global since I will hardly be able to use your code otherwise.

Ok.  I'm still working on this part of it.  Really we need a new Tn5250Win 
object (like Tn5250Field etc.).

> I just compared session.c. I think that your version is probably better 
> but I, as I written in the comments, took the tn5250j project as 
> reference since this one did work correctly on the page I had to make 
> work. I am not a 5250 user so I just fixed the reported error users gave 
> me. :)
>
> Why did you change the type of border_length to unsigned char? Does it 
> have to wrap around 255? It does add a lot of burden to the code.

This could probably be made an int.

> I agree with you it would be better to remove the FCW and FFW values and 
> use variables with significant name. Why do you discard the continuous 
> flags (first, middle, last) and just set your flag to 1? In fact we 
> could use one variable instead of the 3 continued_XXX with an enum (or 
> define).

I think the single FFW variable is fine since according to the docs you 
can only have one field format word for each field.  As for the continuous 
flags, my implementation is not complete and I think we should use yours. 
While what I had in mind for continuous fields would work, what you have 
done better follows the 5250 protocol and thus would make more sense to 
some future coder.

The only change I would make to your implementation is that it assumes 
continuous fields all occur in order one after the other.  The way I did 
cursor progression allows us to use continuous fields even if they are out 
of order.

James Rich

It's not the software that's free; it's you.
        - billyskank on Groklaw