Thanks Jonathan, much appreciated.
Still waiting on access approval to raise an official report for you.
Ian
On Friday, March 20, 2020, 11:38:10 AM GMT, Jonathan Scott
<[email protected]> wrote:
Ref: Your note of Wed, 18 Mar 2020 17:58:43 +0000
Jonathan Scott wrote:
> The problem occurs on the target instruction of a forward
> conditional assembly branch, and appears to suggest that the
> wrong ICTL setting is being used when lookahead processing is
> used in open code to skip ahead to find the target sequence
> symbol.
>
> Major changes to lookahead were implemented for APAR PI90075 in
> January 2018 (to make it compatible with AINSERT) so it's quite
> likely the problem was introduced there (in PTF UI53046 on
> z/OS). Although our HLASM regression testing suite has a large
> number of ICTL tests, it appears that it didn't pick up this
> bug.
Further testing has revealed that this test case fails on every
release of HLASM we have here back to HLASM 1.2 from around 1995.
So the good news for us is that the current team did not break
it, but the bad news is that the original implementation of this
area was in error. I guess not many people use ICTL.
Jonathan Scott, HLASM team
IBM Hursley, UK
