Conference iosg::all-in-1_v30

    G'day,
    
    Re: Base note
    
    I made a mistake in line :-(
    
    > I did a $ EXIT %X10778072 and got no error message just the DCL prompt.
    
      It should have been x=f$message(message_id) I got a No message ....
    
    Anyway I found a solution to the problem it was the customer my cynical
    aproach was correct, they had changed something because DSIN had told
    them !!!!
    
    Reason for the problem and solution :-)
    
    The customer had changed OA$LIB:SMSREORG.COM as follows
    
    $ set file/global_buffer=600 it had been set to 400 BUT did not solve
      the problem so they up'd it.
    
    The files that they had tried to setup the GLOBAL_BUFFER count for were
    
    1. ATTENDEE.DAT
    2. PENDING.DAT
    3. SDAF_(what ever, they have two).DAT
    
    They commented out the line and it worked.  The customer could not tell
    me what the article was about exactly , something to do with memory 
    problems.
    
    Customers sometimes go into a blank phase or the I don't know, or
    better still the Ronnie phase I can't remember !!!!!
    
    I hope that I have helped someone solve a problem.
    
    Sunil
    
    ***************************************************************************
    
    I have two jokes to tell you about the U*X system :-)
    
    1. If any of you have heared about the AWK command you proably know
    that it was called AWK after the initials of the people who wrote it.
    
    Just think if their names had been Stephen, Henry, Ian and Thomas.  It
    would have been very embrassing to tell the customer that they need to
    do a Stephen, Henry, Ian, Thomas on their system !!!!
    
    2. U*X speaking to a Open VMS person.
    
    	U*X person - I never knew that your dog was called 
        SET TERM/NOBROADCAST ?
    
    	Open VMS person ?@#%$$%$^$  (confused)
    
    If you don't know the command for setting your terminal to nobroadcast
    in U*X it's called BIFF !!! Why because the author's dog was called
    Biff !!! Great names for command I am sure that we would all
    immediately think of them !!!!!
    
    Just having a bit of fun.
    

    Sunil,
    
    	The error you saw indicates a $SET syntax error. Normally more
    	information should be returned in a subsidiary error message.
    
    	For example, 
    
		%SET-E-SYNTAX, error parsing '99999'
		-SET-E-VALERR, specified value is out of legal range
    		
    	The first line (%X107710FA) tells you that $SET couldn't cope.
    	The second line (%X10778072) tells you why ...
    
    Cheers,
    
    Iain.

>    
>    The files that they had tried to setup the GLOBAL_BUFFER count for were
>    
>    1. ATTENDEE.DAT
>    2. PENDING.DAT
>    3. SDAF_(what ever, they have two).DAT
>    

  Sunil, be careful with upping global buffers, since it eats up
  memory from the GBLPAGFIL and Global pages. Something about Global
  buffers times bucket size of the data file - there are some STARS
  articles discussing this.

  If the GBLPAGFIL quota gets eaten up, users will experience strange
  problems with apparently missing documents, folders etc.
  If you do a $SEARCH OA$SHARE:OA$DAF_E.DAT blabla.WPL you would get a
  "Unable to map global buffers" and some more errors if this is the
  case.

  Best regards,
  Allan Glargaard who_once_got_blown_by_this
  Digital Service @DMO

    G'day Allan,
    
    Re .3
    
    If that is the case I wonder if the customers were made aware of this in
    the DSIN article.  If not then they should have been made aware.
    
    I am not pointing a finger at anyone especially when I know that the
    customer could not even remember what article they were reading, that's
    a bit of a worry.
    
    I was sure that it was not a syntax error and that the set
    file/global_buffers may have caused this problem.
    
    Thanks for your input
    
    Sunil

RE: .4
>    
>    If that is the case I wonder if the customers were made aware of this in
>    the DSIN article.  If not then they should have been made aware.
>    
>    I am not pointing a finger at anyone especially when I know that the
>    customer could not even remember what article they were reading, that's
>    a bit of a worry.
>    

  Sunil, if you do a search in STARS (DSIN) for "ALL-IN-1 global
  buffer gblpagfil" you get a small page of articles describing
  related problems. They do not all describe in detail how to compute
  the nessesary value of GBLPAGFIL, but article  1152 in the OA1
  database has a good explanation.

  I bet there's a lot of articles suggesting to raise global buffers
  without giving the reader a warning about the requirement for
  a larger GBLPAGFIL parameter - if you find one, you could send a
  comment on the article pointing out that there's a potential danger
  of customer dissatisfaction if the warning is not included.

  If the customer in .0 told us which article it was (s)he read, maybe
  we could at least stop that hole.

  Allan