Difference between revisions of "Covariance through renaming"
m (→Example) |
m (Correction) |
||
(11 intermediate revisions by 4 users not shown) | |||
Line 1: | Line 1: | ||
− | + | [[Category:ECMA]] | |
− | Covariant arguments are part of the Eiffel language, introducing the known problems of CAT calls and global analysis. This page summarizes how to solve a classic covariant problem purely | + | |
+ | Covariant arguments are part of the Eiffel language, introducing the known problems of CAT calls and global analysis. This page summarizes how to solve a classic covariant problem purely through the well-known and understood renaming mechanism in Eiffel. We will see that this solution is more verbose than the covariant solution, but has a set of advantages, including the explicit need to rework the contract, define handlers for CAT calls or the explicit creation of exceptions. | ||
== Example == | == Example == | ||
Line 12: | Line 13: | ||
feature -- Access | feature -- Access | ||
− | last_food: FOOD | + | last_food: detachable FOOD |
feature -- Eating | feature -- Eating | ||
− | eat (f: FOOD) | + | eat (f: FOOD) |
− | + | ||
− | + | ||
do | do | ||
last_food := f | last_food := f | ||
Line 23: | Line 22: | ||
last_food = f | last_food = f | ||
end | end | ||
+ | |||
end | end | ||
</e> | </e> | ||
Line 34: | Line 34: | ||
feature -- Access | feature -- Access | ||
is_vegetarian: BOOLEAN | is_vegetarian: BOOLEAN | ||
+ | |||
end | end | ||
</e> | </e> | ||
Line 50: | Line 51: | ||
feature -- Eating | feature -- Eating | ||
− | eat (g: GRASS) | + | eat (g: GRASS) |
do | do | ||
last_food := g | last_food := g | ||
Line 61: | Line 62: | ||
<e> | <e> | ||
− | + | class | |
GRASS | GRASS | ||
inherit | inherit | ||
FOOD | FOOD | ||
+ | redefine | ||
+ | is_vegetarian | ||
+ | end | ||
+ | |||
+ | feature | ||
+ | is_vegetarian: BOOLEAN = True | ||
− | |||
− | |||
end | end | ||
</e> | </e> | ||
Line 77: | Line 82: | ||
local | local | ||
a_cow: COW | a_cow: COW | ||
− | + | an_animal: ANIMAL | |
do | do | ||
create a_cow | create a_cow | ||
an_animal := a_cow | an_animal := a_cow | ||
− | an_animal.eat (create {FOOD}) -- CAT call | + | an_animal.eat (create {FOOD}) -- CAT call |
end | end | ||
</e> | </e> | ||
− | What is the origin of the CAT call? The error comes from strengthening the precondition of eat through the type system. The precondition in ANIMAL states something like "I will eat everything, as long as it is FOOD", while the precondition in COW states: "I will eat everything, as long as it is GRASS." - so we are | + | What is the origin of the CAT call? The error comes from strengthening the precondition of eat through the type system. The precondition in ANIMAL states something like "I will eat everything, as long as it is FOOD", while the precondition in COW states: "I will eat everything, as long as it is GRASS." - so we are facing a miss-use of the inheritance relation, as a COW is not an ANIMAL - at least not when it comes to food consumption. |
It is a very typical problem of object-oriented modelling, as the subtype relation not only describes a 'COW is an ANIMAL' (observation) but also a 'what we can do to an ANIMAL, we can also do to a COW' (modification) relation. | It is a very typical problem of object-oriented modelling, as the subtype relation not only describes a 'COW is an ANIMAL' (observation) but also a 'what we can do to an ANIMAL, we can also do to a COW' (modification) relation. | ||
Line 106: | Line 111: | ||
feature -- Eating | feature -- Eating | ||
− | eat (g: GRASS) | + | eat (g: GRASS) |
− | + | ||
− | + | ||
do | do | ||
last_food := g | last_food := g | ||
Line 115: | Line 118: | ||
end | end | ||
− | animal_eat (f: FOOD) | + | animal_eat (f: FOOD) |
− | + | ||
− | + | ||
do | do | ||
− | + | if attached {GRASS} f as g then | |
− | + | ||
eat (g) | eat (g) | ||
else | else | ||
− | raise (1,"CAT call") | + | raise (1, "CAT call") |
end | end | ||
end | end | ||
invariant | invariant | ||
− | only_eats_vegetarian: last_food.is_vegetarian | + | only_eats_vegetarian: (attached last_food as food) implies food.is_vegetarian |
end | end | ||
</e> | </e> | ||
− | The advantage of this solution is that it does clearly show that it is possible to get an exception through calling 'animal_eat' in COWs. It does not help preventing the exception from | + | We have renamed 'eat' to 'animal_eat', to create space for a different feature 'eat' that is adequate for cows. We implement 'animal_eat' in terms of 'eat' if the supplied food is the right one. Otherwise, we have the conditions of a CAT call and raise an exception. |
+ | |||
+ | The advantage of this solution is that it does clearly show that it is possible to get an exception through calling 'animal_eat' in COWs. It does not help preventing the exception from occurring, but it shows that there is a deficiency in the code that needs to be taken care of. | ||
== Solution 2: stronger precondition == | == Solution 2: stronger precondition == | ||
Line 143: | Line 145: | ||
feature -- Access | feature -- Access | ||
− | last_food: FOOD | + | last_food: detachable FOOD |
feature -- Status report | feature -- Status report | ||
− | + | likes (f: FOOD): BOOLEAN | |
− | likes (f: FOOD): BOOLEAN | + | |
− | + | ||
− | + | ||
do | do | ||
Result := True | Result := True | ||
Line 155: | Line 154: | ||
feature -- Eating | feature -- Eating | ||
− | eat (f: FOOD) | + | eat (f: FOOD) |
require | require | ||
likes (f) | likes (f) | ||
Line 163: | Line 162: | ||
last_food = f | last_food = f | ||
end | end | ||
+ | |||
end | end | ||
</e> | </e> | ||
Line 180: | Line 180: | ||
feature -- Status report | feature -- Status report | ||
− | likes (f: FOOD): BOOLEAN | + | likes (f: FOOD): BOOLEAN |
− | + | ||
− | + | ||
do | do | ||
− | + | Result := attached {GRASS} f | |
− | Result := | + | |
end | end | ||
feature -- Eating | feature -- Eating | ||
− | eat (g: GRASS) | + | eat (g: GRASS) |
do | do | ||
last_food := g | last_food := g | ||
+ | ensure | ||
+ | last_food = g | ||
end | end | ||
− | animal_eat (f: FOOD) | + | animal_eat (f: FOOD) |
− | + | ||
− | + | ||
do | do | ||
− | + | check attached {GRASS} f as g then | |
− | + | eat (g) | |
+ | end | ||
end | end | ||
invariant | invariant | ||
− | only_eats_vegetarian: last_food | + | only_eats_vegetarian: (attached last_food as food) implies food .is_vegetarian |
end | end | ||
</e> | </e> | ||
Line 211: | Line 209: | ||
From the perspective of the interface, the covariant and the renaming solutions all have the same interface, with the exception that in the renaming cases, COW also has an `animal_eat' command that it inherits from ANIMAL. Though this may seem unwanted, it makes the generation of flattened forms easier and also shows that the 'eat' of ANIMAL and 'eat' of COW are not the same features. | From the perspective of the interface, the covariant and the renaming solutions all have the same interface, with the exception that in the renaming cases, COW also has an `animal_eat' command that it inherits from ANIMAL. Though this may seem unwanted, it makes the generation of flattened forms easier and also shows that the 'eat' of ANIMAL and 'eat' of COW are not the same features. | ||
− | Also, we do not address the problems of CAT calls through genericity and change of export status. These problems need a different treatment (for example | + | Also, we '''do not address the problems of CAT calls through genericity and change of export status'''. These problems need a different treatment (for example through the restriction of the interface through readonly types). |
+ | |||
+ | Interestingly, the approach can be used to model any change of signature, including covariant, contravariant arguments, covariant and contravariant return types and even the change of existing or the introduction of new arguments. It is thus more powerful than the covariant mechanisms available in Eiffel. |
Latest revision as of 12:19, 7 May 2013
Covariant arguments are part of the Eiffel language, introducing the known problems of CAT calls and global analysis. This page summarizes how to solve a classic covariant problem purely through the well-known and understood renaming mechanism in Eiffel. We will see that this solution is more verbose than the covariant solution, but has a set of advantages, including the explicit need to rework the contract, define handlers for CAT calls or the explicit creation of exceptions.
Contents
Example
A class ANIMAL defines a command called 'eat' with an argument 'food':
class ANIMAL feature -- Access last_food: detachable FOOD feature -- Eating eat (f: FOOD) do last_food := f ensure last_food = f end end
And we introduce the food class:
interface class FOOD feature -- Access is_vegetarian: BOOLEAN end
Now we want to model the fact that cows only eat grass, a vegetarian food.
class COW inherit ANIMAL redefine eat end feature -- Eating eat (g: GRASS) do last_food := g end invariant only_eats_vegetarian: last_food.is_vegetarian end
class GRASS inherit FOOD redefine is_vegetarian end feature is_vegetarian: BOOLEAN = True end
We have a potential CAT call, as we can regard a COW as an ANIMAL. An ANIMAL can be fed with any food:
local a_cow: COW an_animal: ANIMAL do create a_cow an_animal := a_cow an_animal.eat (create {FOOD}) -- CAT call end
What is the origin of the CAT call? The error comes from strengthening the precondition of eat through the type system. The precondition in ANIMAL states something like "I will eat everything, as long as it is FOOD", while the precondition in COW states: "I will eat everything, as long as it is GRASS." - so we are facing a miss-use of the inheritance relation, as a COW is not an ANIMAL - at least not when it comes to food consumption.
It is a very typical problem of object-oriented modelling, as the subtype relation not only describes a 'COW is an ANIMAL' (observation) but also a 'what we can do to an ANIMAL, we can also do to a COW' (modification) relation.
Solution 1: explicit exception
We start by clearly stating through the code when a CAT call happens. This solution does not require any further thinking and could be regarded as a flattened form of covariance (though this should be rejected, as this would take away the benefits of making exceptions explicit).
class COW inherit ANIMAL rename eat as animal_eat redefine animal_eat end feature -- Eating eat (g: GRASS) do last_food := g ensure last_food = f end animal_eat (f: FOOD) do if attached {GRASS} f as g then eat (g) else raise (1, "CAT call") end end invariant only_eats_vegetarian: (attached last_food as food) implies food.is_vegetarian end
We have renamed 'eat' to 'animal_eat', to create space for a different feature 'eat' that is adequate for cows. We implement 'animal_eat' in terms of 'eat' if the supplied food is the right one. Otherwise, we have the conditions of a CAT call and raise an exception.
The advantage of this solution is that it does clearly show that it is possible to get an exception through calling 'animal_eat' in COWs. It does not help preventing the exception from occurring, but it shows that there is a deficiency in the code that needs to be taken care of.
Solution 2: stronger precondition
The deeper insight is that ANIMAL is not a correct model of an animal. The contract is too weak: we know from reality that not every animal will eat every food. Animals are picky. We model this by a test predicate that checks if the animal will actually like our food.
class ANIMAL feature -- Access last_food: detachable FOOD feature -- Status report likes (f: FOOD): BOOLEAN do Result := True end feature -- Eating eat (f: FOOD) require likes (f) do last_food := f ensure last_food = f end end
class COW inherit ANIMAL rename eat as animal_eat redefine animal_eat, likes end feature -- Status report likes (f: FOOD): BOOLEAN do Result := attached {GRASS} f end feature -- Eating eat (g: GRASS) do last_food := g ensure last_food = g end animal_eat (f: FOOD) do check attached {GRASS} f as g then eat (g) end end invariant only_eats_vegetarian: (attached last_food as food) implies food .is_vegetarian end
Discussion
From the perspective of the interface, the covariant and the renaming solutions all have the same interface, with the exception that in the renaming cases, COW also has an `animal_eat' command that it inherits from ANIMAL. Though this may seem unwanted, it makes the generation of flattened forms easier and also shows that the 'eat' of ANIMAL and 'eat' of COW are not the same features.
Also, we do not address the problems of CAT calls through genericity and change of export status. These problems need a different treatment (for example through the restriction of the interface through readonly types).
Interestingly, the approach can be used to model any change of signature, including covariant, contravariant arguments, covariant and contravariant return types and even the change of existing or the introduction of new arguments. It is thus more powerful than the covariant mechanisms available in Eiffel.