1. Naming and Binding
A computer system is merely a bunch of resources that are glued together with names
Thus, much of system design is merely making choices regarding naming
three abstractions in a system
memories store named objects
interpreters (processors and translators) manipulate named objects
communication links connect named entities

2. Naming and Binding
binding - assignment of name, value, or other attribute to an object
the name can be symbolic, binary address, ...
the value can be text, binary value, ...
value
name

3. Naming and Binding naming
syntax: what are the rules in constructing a valid name?
range: how many different objects can you name?
order: binary numbers are ordered, add offset to one to construct another
granularity: what size unit is assigned a name - bit, byte, word, block?
homogeneity: array name is name of collection of homogeneous data, while struct/record name is name of collection of heterogeneous data

4. Naming and Binding naming (cont’d)
dimensionality: vector, array, etc.? (true segmentation uses a two-dimensional name)
context-dependent / overloaded name: abbreviated name must use some form of context to disambiguate; that is, a name resolution mechanism to resolve the name among multiple possible object references; context search rules typically called "scoping")
aliases: multiple names for same object
protection: do you associate access rights with names or with objects? (e.g., consider file system)

5. Naming and Binding binding type static binding - done once
dynamic binding - done several times (dynamic binding of names usually requires a naming table)
binding time - when done - e.g., compile time, run time
early binding
late binding, sometimes called lazy evaluation - done only when needed (e.g., "just-in-time" compiler)
static, early binding - e.g., compile-time binding in a strongly-typed language, catch errors early, can execute efficiently

6. Naming and Binding binding time (cont’d)
dynamic, late binding - e.g., run-time binding in an interpretative language, flexible, easier to debug, but often slow

7. Naming and Binding dynamic assignment of value dynamic naming
static binding
dynamic binding
assemble time
symbol name to address
link time
external addr. resol. relocation
load time
relocation
run time (can be sub-divided, e.g. procedure-entry time)
“dynamic” linking
is really a misnomer, since binding done at run time on first call – this is lazy evaluation
value to mem location
parameter passing
I/O buffers
choosing ISR (through int. vect. table)
segmentation/paging (thru seg/pg table)
file system directory
file index nodes
linked list with dyn. memory allocation
dynamic
assignment of
value
dynamic naming
(separate name
table)
name table is in
links
main advantage is speed (low overhead)
main advantage is flexibility (ability to change named object without modifying the invoker of the name)

8. Naming and Binding quote attributed to David Wheeler of EDSAC fame
"All problems in computer science can be solved by another level of indirection."
also attributed to Butler Lampson
Kevlin Henney corollary: "...except for the problem of too many layers of indirection."