Persistence predicates

Research: This page describes research about Eiffel, not the actual language specification.

Introduction

Using STORABLEs in an application has at least two significant disadvantages.

One is that using them for long-term persistence is vulnerable to changes in the class.

Another is that too much data may be stored, resulting in large files (or database occupancy, or whatever) and slow reload times. Some data, such as internal caches, are probably best not serialized at all. For other data, it may be better just to store a shorthand identifier to enable the data to be rebuilt (maybe on demand) at retrieval time.

This article explores ways to address the second problem. The first problem is best tackled by only using STORABLEs as a medium-term caching mechanism.

Transient data

Java tackles this problem with the keyword transient. An attribute so marked is not stored at serialization time. When the object is retrieved the attribute will be void.

This seems rather limited to me. It can't be used if the class invariant constrains the attribute to be non-void. And I don't think there is a way for a descendant class to override this behaviour (it's a long time since I've written any Java, so I could be very wrong here).

My thinking is that there should be a way to specify whether or not a particular attribute should be stored on any given occaision, and a way to specify what to do at retrieval time.

Persistence predicates

A persistence predicate is a routine of type PREDICATE [ANY, TUPLE [!<attribute_type>]] associated with an attribute of type <attribute_type>. At storage time, the predicate will be called with the instance of the attribute as its argument (if the attribute is non-Void). If it returns True then the attribute is stored. If it returns False then the attribute is not stored, but it may be re-created at retrieval time.

Restoration routines

If a persistence predicate returns False, it may be that the programmer does not need this data at all (such as the case of an internal cache). But in some cases the programmer will want the data to be recreated at restoration time (this may be necessary to restore the class invariant), rather than retrieved from the persistence mechanism.

Take an example of a class which has an attribute named `gadget' of type GADGET_FROM_DATABASE having an attribute named `id' of type INTEGER. The class invariant requires that`gadget' is not Void, but the size required to store `gadget' is very large. The class also has a secret routine `create_gadget_from_identifier (a_id: INTEGER) which can be used to create a GADGET_FROM_DATABASE given its `id' value. In such a case, rather than store `gadget' it is probably better to just store the `id' value, and then call `create_gadget_from_identifier' at retrieval time.

A restoration routine is similar to a creation procedure, in as much as it cannot reply on the class invariant holding. It can rely on attributes for which the persistence predicate evaluates to True as having been restored, though.

Syntax

One new keyword restore will be required. This can occur in several places:

  • After the create clause in the class prolog. What follows is a list of restoration routines, just like the list of creation procedures. Typically this will occur in the form:
    restore {STORABLE}
     recreate_attributes
  • After the attribute keyword and optional body. What follows is:

    1. The tag persistence_predicate: followed by one of:

    a) True meaning always persist.

    b) False meaning never persist.

    c) The keyword agent followed by a declaration of a PREDICATE.

    2. An optional body (introduced by the tag body:) like the attribute body, and this is code that is executed only if the attribute was not persisted. These routines run after all persisted attributes have been restored, but before the restoration procedure is run. It is only used for initialization of attributes that do not depend upon other restoration attribute bodies having been run.

  • Within a restoration routine as an instruction. Here it is used analagously to a create instruction, to run one of the named restoration routines specified in the class of one of the attributes that was not persisted.

Example

class A
 
create
 make, make_from_db
 
restore {STORABLE}
 initialize_attributes, recreate_attributes
 
feature {NONE} -- Initialization
 
 make
   -- Initialize `Current' with a new gadget.
  do
   initialize_attributes
   gadget.save_to_database
   -- other stuff ...
  end
 
 make_from_db (a_gadget_id: INTEGER; ...)
   -- Initialize `Current' using `a_gadget_id' to create `gadget' from database.
  do
    recreate_attributes (a_gadget_id, ...)
    -- other stuff
  end
 
 initialize_attributes
   -- Initialize `gadget' and ...
  do
   create gadget.make_new
   -- other stuff ...
  end
 
 recreate_attributes (a_gadget_id: INTEGER; ...)
   -- Recreate certain attributes.
  do
   create_gadget_from_identifier (a_gadget_id)
   -- other stuff
  end
 
feature -- Access
 
 gadget: GADGET
   -- Gadget retrieved from DB via its `id' attribute
  attribute
  restore
   persistance_predicate: agent is_gadget_persisted (attribute.id) -- `True' and `False' would also be acceptable values, 
                                                                   --  with `True' being the default
   body: agent create_gadget_from_identifier (attribute.id)
  end
 
end
 
class B
 
inherit
 
 A
  redefine
   make
  rename
   gadget as gadget_from_a
  end
 
create
 
 make
 
restore {STORABLE}
 
 restore_gadgets
 
feature {NONE} -- Initialization
 
 make
   -- Initialize `Current' with a new gadget.
  do
   Precursor
   create gadget.make_new
  end
 
 restore_gadgets (a_id, a_other_id: INTEGER; ...)
   -- Restore `gadget' and `gadget_from_a' from database.
  do
   restore recreate_attributes (a_other_id, ...)
   create gadget.make_from_databse (a_id)
  end
 
feature -- Access
 
 gadget: MY_GADGET
   -- Gadget createable from database.
  attribute
  restore
   persistance_predicate: False
  end
 
end

Here the ECMA attribute keyword is reused to denote the instance of `gadget'. This is only available within the attribute restore clause, not in a restoration routine.

Now here are some scenarios in which objects of classes A and B are stored:

local
 a: A
 b: B
do
 create a.make
 create b.make
 
 -- Save a to a storable, using defaults for retrieval
 a.independent_store (a_medium)
 
 -- Assuming `a.is_gadget_persisted (a.gadget.id)' returns `False', `a.gadget' will not be saved, but `a.gadget.id' will be saved
 --  for use in the local (attribute-specified) restoration routine `create_gadget_from_identifier.
 
 -- Save b to a storable specifying a restoration routine
 b.independent_store (a_medium, agent b.restore_gadgets (b.gadget.id, b.gadget_from_a.id, ...))
 
 -- Assuming `b.is_gadget_persisted (b.gadget_from_a.id)' returns `False', then neither gadget will be stored, but their identifiers
 --  will be recorded so they can be passed to `restore_gadgets' at retrieval time.

Redefinition

The open/closed principle is honoured (I think). The author of the class can determine that an attribute must always be stored (this is the default, and is the current situation), or that it is never stored (for a pure cache attribute), or provide flexibility by specifying an agent.

The agent's routine could be deferred, frozen (yuk!) or just a normal routine, in which case descendant classes can redefine it.

The actual instance data that is passed to the persistance_predicate and the restoration_routine could be specified via an agent, which allows for further flexibility. For restoration, it might even involve an EV_DIALOG (heaven forfend)!

Class correctness conditions

The possibility of the persistance_predicate evaluating to (or statically specified as) False on an attribute that is required to be non-Void (or non-default for an expanded attribute) implies that either inline restoration code must be supplied, or the restoration routine has an appropriate postcondition clause.