Notes from z390 July developer meeting

13 views

Skip to first unread message

Abe Kornelis

unread,

Jul 8, 2026, 5:25:00 PMJul 8

to z390dev...@googlegroups.com, z390

Hi all,

just a short update.

We met with John Ganci, Abe Kornelis, Bob Schulien, and Larry Simmons.

We mentioned progress on zCobol support: STRING and UNSTRING
verb support has been considerably improved. Improvements for
WRITE, SEARCH, and GO TO DEPENDING ON are progressing slowly.

Abe mentioned he's been working to get all the test programs in the
repository to become part of the automated regression test suite.

He also mentioned zCics and Guam test cases having an issue
that he could not resolve in the time available. To be resolved
at some later point in time.

We spoke about Java code reformatting at length. John's been
working to improve the code formatting for two of the smaller
java programs we have. Abe has created a prototype linter that
is capable of enforcing rules. We agreed to work together to do
both the reformatting and the linter definitions in close cooperation.

John brought up the issue of modal instructions. He recently fixed
modal issues for TRT, TRTR, ED, and EDMK. Many other instructions
also have issues with handling addresses in 64-bit mode.
We decided to tackle the issue in small increments dealing with
just a few instructions at a time. We shortly discussed an option
to add DAT to z390 so we can use virtual storage concepts to
implement 64-bit storage on a memory array that has only room
for (almost) 2**31 bytes.

Next we discussed our Jave upgrade. Currently z390 still depends
on Java v8 - or a newer version operating in v8 compatibility mode.
Current releases of Java are 21 and 25. Abe promised to do some
investigation so we can return to this topic next month.

Abe shortly mentioned progress on the project to reconcile the
two diverged versions of the zVSAM V2 design documents.

Then we discussed the floating point registers. z390 implements
the floating point in an array, with three shadow arrays holding
converted values, one array for each type of floating point number.
To implement Vector Registers, we need to re-implement the
floating point registers, as they overlap with a subsection of the
Vector Registers. It is going to be a major effort to make that
change in z390.

Next we decided to create a new distribution, which will come
out in a day or two. Abe to prepare V1.8.4.4 asap.

Finally, we discussed a pending PR that suppresses the full
10-minute regression test run on the github server when a
change contains only doc changes.

We spent a few words on the IBM announcement for the new z17.
Since no new PoP is expected this year, this has no impact for z390.