Number: 718

Date: 16-Apr-84 13':48':07

Submitter: le.pasa

Source: Bill White (Teknowledge)-Raim.pasa

Subject: Need to document folklore on how to avoid running out of array space

Lisp Version: Fugue.6

Description: '
User needs to know what folklore is avaialble for not running out of array space as a result of using system software.  User is not creating his own windows.  He is using Dedit, Tedit, Chat, etc.  He needs to know ways of releasing space (e.g. RESETDEDIT) to avoid running out of array space.'
'
-----'
'
Sender': Sannella.PA'
Date': 13 Apr 84 13':29':04 PST (Friday)'
Subject': AR 692': Gets Arrays full after using EXPANDBITMAP a lot'
To': Kiewiet.pasa'
cc': LispSupport.pa, Burton.pa'
From': LispSupport.pa'
Reply-To': LispSupport.pa'
'
I have submitted this as AR 692, even though I suspect that it may just be the user doing something wrong.  I suggest you send back the following':'
'
1) "Arrays Full".  Are you sure that your program is not holding pointers to old bitmaps after you use them (directly or indirectly)?  RECLAIM will not reclaim old bitmap space if you hold pointers to them.  It will also not reclaim the space if you construct circular structures.'
'
2) "I need lower level control ... to bitmaps".  What do you need to do that you can''t do with BITMAPBIT?'
'
Richard --- do you have anything to add?'
'
-----'
'
Date': 18 Apr 84 21':54 PST'
From': Kaplan.pa'
Subject': Re': Ar#716'
In-reply-to': vanMelle.pa''s message of 16 Apr 84 14':04 PST'
To': le.pasa, Lispsupport'
cc': vanMelle.pa, 1100Support.pasa, JonL.pa'
'
I think Bill''s response to AR 716 (arrays full) is the correct one.  If his arrays are not being collected, then it must be because he is keeping a pointer to them somewhere.  In this situation there are 2 ways in which the system could be considered to be "at fault"'
'
1.  He is not really keeping an explicit pointer to the bitmap--rather, he has some circular data structure that is also pointing to the bitmap or array.  The system currently does not collect circular structures.'
'
2.  Arrayspace is too small.  This could be improved by allowing arrays to be allocated out of main data space (I''ll submit an AR for this), and also be expanding the address space up to 24 bits (an AR for this already exists).'
'
I think he should follow Bill''s suggestion and try to find out whether his own code is at fault for keeping pointers to bitmaps that he doesn''t really intend to reference again.'
'
A separate AR should be submitted (Attn Burton ?) for lower level access to bitmaps.  What does he really want, if BITMAPBIT is insufficient?'
'
--Ron'
'
-----'
'
Date': Thu, 19 Apr 84 11':31 PST'
From': Raim.pasa'
Subject': Re': Ar#716'
In-reply-to': "Kaplan.pa''s message of 18 Apr 84 21':54 PST"'
To': Kaplan.pa'
cc': le, Lispsupport.pa, vanMelle.pa, 1100Support, JonL.pa'
'
Ron,'
'
Thank you for your response.  What''s sorely lacking down here is the folklore necessary to help a user explore reasons for unwanted behaviors.'
'
Bill''s suggestion was ': "...run STORAGE and try the GCHAX library package for tracking down who is at fault...." '
'
Glancing at the description of STORAGE in the ILM reveals that its output is implementation dependent.  Could you please explain what the current implementation of STORAGE prints out.  (I''ve run it, but am uncertain as to what it all means.)'
'
As for GCHAX,  no user documentation exists except for a message from VanMelle describing a new function, \SHOWGC.  I presume something like (\SHOWGC T ''BITMAP) is what''s needed.  Is there a more complete description of GCHAX somewhere?'
'
It would be nice for us to be able to give our many users who are running into the Arrays full problem a more definitive means of tracking its cause within their own code.'
'
Could you briefly outline how YOU would use STORAGE and GCHAX  to track down this problem?'
'
Thanks,'
'
--Marty'
'
-----'
'
Date': 19 Apr 84 12':45 PST'
From': vanMelle.pa'
Subject': Re': Ar#716'
In-reply-to': Raim.pasa''s message of Thu, 19 Apr 84 11':31 PST'
To': Raim.pasa'
cc': Kaplan.pa, le.pasa, Lispsupport.pa, 1100Support.pasa, JonL.pa'
'
Try <Lisp>Library>GCHAX.press'
'


Workaround: 

Test Case: 

Edit-By: Sannella.PA

Edit-Date: 19-Apr-84 16':00':37

Attn: VanMelle.pa, Raim

Assigned To: 

In/By: 

Disposition: 

System: Operating System

Subsystem: Virtual Memory

Machine: 

Disk: 

Microcode Version: 

Memory Size: 

File Server: 

Server Software Version: 

Difficulty: Moderate

Frequency: Intermittent

Impact: Moderate

Priority: Hopefully

Status: Wish

Problem Type: Documentation

Source Files: