Re: Will IBM ever expand system object names?

Hi Jon,

I suspect that AI would be especially useful at helping to write the code
of an OS. Just saying, the investment is much lower now than in the past.

-Steve


On Wed, May 29, 2024 at 5:20 PM Jon Paris <jon.paris@xxxxxxxxxxxxxx> wrote:

Most things are "easy" if you throw enough money at it Steve.

But for what ? I don't see any payback for the investment.


Jon P.

On May 29, 2024, at 5:09 PM, Steve Richter <stephenrichter@xxxxxxxxx>

wrote:

On Wed, May 29, 2024 at 3:26 PM Jon Paris <jon.paris@xxxxxxxxxxxxxx>

wrote:

Where would you store the relationship between the 10 char name the

system

would have to create and the long name? Would you allow calling by the
short name? That would match the way SQL names work. Either way it slows
down calls.

Regarding the slowing down of calls, simply flip the switch that enables
all the power 10 cores to run ibm i workload.

Just to give a tiny example of the work needed. Think of all the APIs
that accept a 20ch compound name. How big are you going to make the new
names? 30? 40? They can't be arbitrary because such APIs have to split

the

components. Unless you include a separator? Now you have extra logic to
test for its existence. Then there are edge cases such a 20 char field

with

a five char library name and a 14 long object. even the length of the
field doesn't tell you how to process it.

"easy". Implement javascript like dynamic objects, strings and garbage
collection into the OS. Write new versions of the system APIs which
accept dynamic objects as parameters. Use AI and copilot to convert the
system APIs to the new version.

-Steve