Number: 29

Date: 16-Mar-84 23':58':38

Submitter: Sannella.pa

Source: ROACH.PA

Subject: ACCESSFNS of system records not saved on SYSRECORDS

Assigned To: 

Attn: vanM, Kapl, Burton

Status: Open

In/By: 

Problem Type: Bug

Impact: Annoying

Difficulty: Easy

Frequency: Everytime

Priority: Perhaps

System: Programming Environment

Subsystem: Record Package

Machine: 

Disk: 

Lisp Version: 

Source Files: 

Microcode Version: 

Memory Size: 

File Server: 

Server Software Version: 

Disposition: 

Description: '
Date': 12 MAR 84 17':38 PST'
From': ROACH.PA'
Subject': Missing STREAM accessfns'
To':   LISPSUPPORT'
cc':   ROACH'
'
     ACCESSFNS for datatype STREAM are missing in the declaration'
of STREAM found on SYSTEMRECLST.  They are also missing in EXPORTS.ALL.'
     Responsible person': KAPLAN.'
				Kelly'
'
-----'
'
Date':  7 Apr 84 15':29 PST'
From': Kaplan.pa'
Subject': AR 29':  Miising accessfns for STREAM'
To': Lispsupport, Vanmelle'
'
This is not a problem with STREAMs.  It seems to be a problem with all sysem datatype declarations':  The function SAVEONSYSRECLST on RECORD only includes the fields of the datatype, not the tail of the declaration, which is where the accessfns would be.'
'
I don''t know who decided to do it this way or why, but Bill was the last editor of this function.'
'
Not understanding the issues, I won''t work on this further.  Either it should be passed on to Bill, or marked as declined, I guess.'
'
--Ron'
'
-----'
'
Date':  7 Apr 84 15':46 PST'
From': vanMelle.pa'
Subject': Re': AR 29':  Miising accessfns for STREAM'
In-reply-to': Kaplan.pa''s message of 7 Apr 84 15':29 PST'
To': Kaplan.pa'
cc': Lispsupport.pa, vanMelle, Burton.pa'
'
One of the reasons that SYSRECORDS does not save the tail of the declaration is that the inspector doesn''t use it.  I assume it was Richard''s decision that inspect of a system record included only the "real" fields, not any accessfns or other overlays; but in any case, I agree with it.  In fact, I often wish this were true for user records as well; in most cases I have seen, the additional fields are just synonyms or alternative perspectives on the main record, and I find it confusing and/or annoying to find them in the inspect window.  Once in a while, the accessfns even have undesirable side effects when accessed by the inspector.'
'
	Bill'
'
-----'
'
Date':  9 APR 84 19':36 PST'
From': JONL.PA'
Subject': Re': AR 29':  Miising accessfns for STREAM'
To':   vanMelle, Kaplan'
cc':   Lispsupport, Burton, JONL'
'
In response to the message sent   7 Apr 84 15':46 PST from vanMelle.pa'
'
Before deleting the inspector facility that shows all "fields" (including'
the ACCESSFNS ones), be sure that there is some way to get that effect'
when wanted.  For example, I was inspecting some random datatype (maybe it'
was the PILOTBBT?) and the basic fields where something like BASEHI and BASELO,'
which when printed out just look like two random numbers; but seeing the BASE'
"field" make it all clear -- BASE just did a \VAG2 of the appropriate other'
fields.'
'
-----'
'
From': KAPLAN.pa'
Date': 10-Apr-84  7':41':16 PST'
Subject': Re': AR 29':  Miising accessfns for STREAM'
In-reply-to': JONL''s message of 9 APR 84 19':36 PST'
To': JONL'
cc': vanMelle, Kaplan, Lispsupport, Burton'
'
Actually, I also question the wisdom of suppressing accessfns in the inspector. Sometimes that''s the only way of giving a symbolic interpretatation to what would otherwise be garbage.  The ACCESS field of a stream is one example that comes to mind.'
'
Seems to me that the accessfns ought to at last be passed thru on sysemreclst, ad perhaps then the inspector could have a flag that suppressed them or not.'
'
--Ron'


Workaround: 

Test Case: 

Edit-By: Burton.pa

Edit-Date: 16-May-84 11':38':40