On 7/16/2014 9:41 AM, Tom Marchant wrote:
On Wed, 16 Jul 2014 08:52:32 -0600, Steve Comstock wrote:
Actually, some improvements / corrections are included below
Data_in ds PL3
Data_out ds CL7
* Edpat dc x'40202020202060'
Edpat dc x'40202020212060'
~~
Then your code might look something like this:
mvc Data_out,Edpat
* la 1,Data_out+4
la R1,Data_out+5
~
edmk Data_out,Data_in
cli Data_out+6,c'-'
* je positive
jne positive
~~~
bctr R1,0
mvi 0(R1),c'-'
mvi Data_out+6,c' '
positive ds 0h
Steve,
Data_out+6 will always be c'-'. The only values that have any significance for
EDIT and
EDMK are X'20', X'21' and X'22'. Any other value is a Message byte and will be
left
undisturbed with the edit mask that you have defined. With your initial edit
mask it might
have been replaced with the fill character.
The first significant character (non-leading zero or zero
after significant start), turns on the s-trigger. The presence
of a positive sign in the last data byte will turn the s-trigger
off and so any remaining message characters will be replaced
by the fill character.
The two edit patterns work the same, of course, except when
there are four leading zeros; in that case the s-trigger is
not turned on (thus not setting the R1 value correctly).
I no longer have access to a system to test / demonstrate / verify
this, but I believe it to be true.
-S
