On Mon, 7 Apr 2014 22:03:40 -0300, Clark Morris <[email protected]> 
wrote:

>On 7 Apr 2014 10:41:50 -0700, Bob Shannon wrote:
>
>>>Why is it that SYS1.NUCLEUS can't be a PDSE? SYS1.LINKLIB?
>>>SYS1.LPALIB, SYS1.PARMLIB?  Could it be the short sighted and inadequate 
>>>implementation?
>>
>>Common Clark. You should already know that NIP is too early in the IPL 
>>process to use PDSEs.
>
>I doubt that anyone would object to a 10 megabyte IPL text (and 1
>should be sufficient) that would know what to do with a PDSE.

I don't know how much code is required to properly support PDSE, and I doubt 
that you do. 
Please don't suggest that the IPL Text should contain some subset of the code 
required to 
fully support PDSE. Something is bound to break if that is done.

IPL Text goes on Cylinder 0, track 0 and is limited to about 53K bytes. Could 
that be 
architected differently? Sure. How many tracks should it occupy? What happens 
if the code 
required grows beyond what is allocated for the IPL text? If you suggest that 
it go into a 
data set, then you have the same problem: that data set can't be a PDSE.

PDSE has has a lot of changes over the years. I wouldn't want anything added to 
the IPL Text 
unless it is rock solid.

Finally, why is it so important that NIP should support PDSE? And how much is 
it worth to make 
it so that it can? The process of initializing MVS is a multiple stage process 
and it works well. 
Over the years there have been other things that people have complained that 
NIP can't do. 
Support user catalogs, for example. But there is no business case for making 
the large changes 
that would be required.

>I find it ludicrous that the code that
>runs a vital service is a started task.

Why is it ludicrous? MVS has always had many of its parts implemented in 
started tasks. Even 
*MASTER* is a started task. Unix is architected similarly. When you write an 
operating system, 
you can architect it any way you choose. And people will come along and 
criticize your 
decisions, too.

>...  it makes no sense to not have the code needed to read PDSEs
>available at NIP and IPL time.

Why do YOU need PDSE support at IPL time? 
As far as I know, it makes no sense to  _require_  that PDSE support be 
available at IPL time. 
If you have a business case, or a requirement for your code, please state it.

>For those who think 1 megabyte is a large amount of space ....

The issue isn't the amount of space required, except for the currently 
architected limit to 
the size of the IPL text.

-- 
Tom Marchant

----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN

Reply via email to