[net.lang] Static links vs the Display

roy@gatech.UUCP (10/20/83)

The trade-off made by using a display instead of static links is
one of time.  Static links require an amount of time proportional
to the lexical distance between the level on which the reference
was made and the level on which the variable was defined.  To
access a variable one level away requires one to follow the static
chain one level, access to a variable two levels away requires one
to follow the static chain two levels, etc.  Using a display makes
access to variables at all levels equally difficult, one must look
in the display for the base address for a lexical level, and then
use that address and an offset to access the variable.  However,
procedure entry/exit time is increased by the time needed to update
the display.

It is really a matter of whether you want to pay on a variable by
variable basis (static link) or on a procedure basis (display).
-- 
"Roy J. Mongiovi"
  USA:  School of ICS, Georgia Institute of Technology.  Atlanta, GA 30332
CSNet:  Roy@GaTech                           ARPA:   Roy.GaTech@UDel-Relay
 uucp:  ...!{sb1,allegra,ut-ngp}!gatech!roy  ...!duke!mcnc!msdc!gatech!roy