CONTACTS 

SUBJECTS
Archiving Diary Records, and SAA
Diary Delete or Update

0504 -    ..
0505 - Summary/Objective
0506 -
050601 - Encountered possible error in code created at ref SDS 1 line 35.
050602 -
050603 - Modified code to see if it solves problem.
050604 -
050605 - Discovered problem was likely caused by routine that deletes records,
050606 - discussed below.
050607 -
050608 -
050609 -  ..
050610 -
0507 -
0508 -
0509 - Progress
0510 -
051001 - Last night I found there were four blank lines in the weekly poiner
051002 - file at ref OF 1 line 2.  They were on lines 2 - 5, and it appeared
051003 - the thing was not properly sorted, although I did not study that
051004 - carefully.
051005 -
051006 -     I ran the routine in debug mode, and it worked correctly.  I
051007 -     re-constructed the current "00" directory using the saved schedule
051008 -     "00" file from the day before, and gluing it to the next day for
051009 -     the current schedule.  Put it in directory "09" so I could quickly
051010 -     jump it back and forth for testing.  However, multiple tests of
051011 -     compiled code could not duplicate the error.
051012 -
051013 - Today, I added code to remove any blank lines in the pointer files
051014 - during the operation to sort them.  I entered the explanation in the
051015 - file where the original work was performed at ref SDS 1 line 110.
051016 -
051017 - Initially I considered only doing the remove blank line op for 04702,
051018 - since that is the only place were new data is entered, but actually
051019 - it is needed for both 04702 and 0702051 that updates pointers.
051020 -
051021 -     The work done on macro 070201, likely means the change made here
051022 -     is not needed.  Normally there should never be any blank lines in
051023 -     the weekly pointer files.  None are created so none should occur.
051024 -     It is possible though for someone to open the files to examine
051025 -     them, and they could inadvertantly add blank lines.  So, this
051026 -     change may be helpful for that.
051027 -
051028 -
051029 -  ..
0511 -
0512 -
0513 - 070201
0514 -
051401 - Line 184 - * Del d:\sd\08\, this code uses Supersort to remove entries
051402 - for a record being deleted.
051403 -
051404 - The record length was only 256.  I changed it to 400.  This may have
051405 - been a major cause of the prior problems.  Could be we did not really
051406 - need to create the other code, but since it is faster and takes less
051407 - code, I will keep it.
051408 -
051409 - This code still keeps the same filename for sort input and output
051410 - file, which was considered the cause of the prior problems.  As of
051411 - today, I am revising that analysis.  I suspect the problem all along
051412 - has been the 256 record length in this routine.  Since it is rare that
051413 - records are deleted, that accounts for the rarity of weekly pointer
051414 - files getting messed up, per analysis on 940303. ref SDS 1 0184
051415 -
051416 -
0515 -
0516 -