LSTSRV-L Archives

LISTSERV Site Administrators' Forum

LSTSRV-L

Options: Use Monospaced Font
Show Text Part by Default
Show All Mail Headers

Topic: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Eric Thomas <[log in to unmask]>
Thu, 25 Nov 1993 16:37:33 +0100
text/plain (54 lines)
On Thu, 25 Nov 1993 07:36:17 -0800 John Bacon <[log in to unmask]> said:
 
>The GLOBALVs are stored in memory and the disk image is updated from the
>memory copy whenever any change in their values are made.
 
I don't know  who you got this  from but it must be  a very superstitious
person :-) The only time disk is  ever updated from memory is when you do
a GLOBALV  INIT: the disk  file is loaded into  memory and, if  there are
superfluous entries, it  is rewritten from memory. The rest  of the time,
GLOBALV just  appends new  records, and then  only if you  do a  SETP (as
opposed to SET).
 
>Even worse with the present file  systems only blocks which have changed
>in the memory copy might be overwritten.
 
I don't  see how the  file system  implementation would have  any effect.
This  is certainly  not possible  with  a RECFM  V file  where you  can't
rewrite  "just the  records that  have changed"  (unless the  replacement
records have exactly the same length,  or you want to truncate the file).
When  it rewrites  the disk  file,  GLOBALV makes  a new  disk file  from
scratch.
 
The reason it  is not a good  idea to edit LASTING GLOBALV  is that XEDIT
strips all  trailing blanks from  variable files  whether you like  it or
not.  If you  have  variables with  a  trailing blank  this  may cause  a
problem. You  can use GLOBALV  SETP to  delete the variable  safely. This
being said, LDBASE XYZVM is the best solution to the original problem :-)
 
  Eric
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
>#CP LOGOFF. The LOGOFF without CMS processing or traps is important. (Re
>IPLing CMS from CP works too.)
>
>--
>John Bacon [log in to unmask]

ATOM RSS1 RSS2