Re: [LINUX5250] Cursor positioning

[MarkB@xxxxxxxxxx]

UPDATE: To see an example of the difference between tn5250 and Client
Access:
1. Start SEU
2. Do a "find <sometext>"  where <sometext> is in the text you are
editing/displaying
The cursor's home position should now be the location of the find result.
3. move the cursor and press home
(Client access) cursor ends up in last find result position.
(tn5250) cursor ends up in SEU's command field.

This is because Client Access is honoring the home position set by the IC
order, whereas tn5250 won't honor it because it checks pending_insert
which will be false.

My fix isn't a fix:
I was browsing through CVS and realized that what I suggested here to fix
the "ignoring home" problem, is actually an undoing of a (partial?) fix for
"multiple WTD" problems.

The diff tells all.
http://cvs.sourceforge.net/cgi-bin/viewcvs.cgi/tn5250/tn5250/src/display.c.diff?r1=1.33.2.2&r2=1.33.2.1
 "Second patch from Dave McKenzie fixing cursor positioning issues when the
  emulator is sent more than one WTD." - diff comment

In any case, without my "fix" below, things are broken. tn5250 does not
behave the way Client Access behaves.
With my "fix", I've seen it work with multiple WTD orders and mupliple ICs.
Maybe the problem is specific combinations of multiple WTDs AND multiple IC
orders.
One issue that comes up with this fix, is that in general the most you have
to type home to invoke it as an AID key should be twice.
1. -->to the home position if it's not already there.
2. -->AID key.
Somehow, with this fix, you have to end up pressing it 3 times. This might
be because the check to see if Home should be processed as an aid key is
done at the end of tn5250_display_kf_home.

I am working to understand these issues. Any input is appreciated.

Thanks,
Mark

email
markb
cms400
com