Deleted Key errors

[Howard Wolowitz]

From: "John Esak" <john at valar.com>
Subject: RE: Deleted Key errors


>
>> I know the code for this particular program does some weird stuff
>> but it is
>> in production and I can't fix it until some quiet time.
>>
>> For example: When the user saves a record the processing uses the
>> "escape"
>> command to save the current record and move on to the next free
>> record.  I
>> have no idea how that works but I do know that it stopped working when I
>> removed the escape and used a write command instead.
>>
>> ai="" ; ba=1; gosub qoh; escape ;  break on  ; screen, 1;goto top
>>
>
> Don't know why I'm jumping in here... but I can't help it.  Why would you
> expect the code *after* the ESCAPE to work?  Just curious. If this is in a
> @wlf or @wef process, the ESCAPE is one of the designated "ending" 
> commands.
> But I know you *know* this... so, if it is in INPUT rocessing, are you 
> sure
> this is where the processing pointer *is*?  In other words have you put a
> debug after the ESCAPE to ensure that the things after the ESCAPE are
> actually happening?  Is it possible the ESCAPE  functions and processing 
> is
> really picking up somewhere else?  For example... if you have already
> pressed ESC (meaning the user has pressed it physically) and you then hit
> this ESCAPE in your processing... what would you expect it to "do"?. I 
> guess
> I'm asking "where" did this processing get activated and how?  Could it be
> in an @wuk?
>
> The gist of my question is if you are not in @wlf/@wef where ESCAPE ends
> things... and if you are not in INPUT where ESCAPE has *already* been
> pressed?  how did you *get* to this code and what do you want the ESCAPE 
> to
> do, because if as you asay you want to be brought to the next record...
> wouldn't you want an END here in stead of an ESCAPE?

The code was written eons ago and has been working user SCO Unix.

The escape I mentioned is after the user already pressed escape to save the 
record - when filepro would start at the top of the processing table. 
That's why it makes no sense to me.  The program was in add records mode and 
the "escape; screen; goto top" seems to save the current record and move to 
the next free one.

You are right that "end"  (perhaps "write; end") is the proper way to do it 
and that is what I will be testing when no one is using the system and I'm 
still awake.

FYI  Tom has reported that since we moved to Linux we are getting 
"Rebuilding Free Chain" and numerous files.  We are compiling a list and 
will be looking for a common technique to blame before I report it to 
fptech.

Has anyone else been plagued with free chain errors on Linux?

Howie