mirrors/eiffel-org
1
0
Fork 0
mirror of https://github.com/EiffelSoftware/eiffel-org.git synced 2025-12-06 14:52:03 +01:00
Code Issues Packages Projects Releases Wiki Activity

Update wikipage EiffelBase, The Kernel. (Signed-off-by:tqa7ve2mnbntqnfca3i6rk7arhc2kxr8).

Browse Source 
git-svn-id: https://svn.eiffel.com/eiffel-org/trunk@1814 abb3cda0-5349-4a8f-a601-0c33ac3a8c38
This commit is contained in:
eiffel-org
2017-05-07 21:30:30 +00:00
parent a8336d7783
commit 2e4ec3e4a6
1 changed files with 5 additions and 471 deletions
Download Patch File Download Diff File Expand all files Collapse all files

476
documentation/17.01/solutions/basic-computing/eiffelbase/eiffelbase-tutorial/eiffelbase-kernel/index.wiki
View File

@@ -2,485 +2,19 @@
[[Property:weight|0]]
[[Property:uuid|A26C5BCA-20A5-4EE6-9710-F4E8B72DBA10]]
In addition to basic concepts close to the language level, the Kernel covers such common needs as '''input''' and '''output''', '''storage''' and '''retrieval''' of objects on persistent storage, fine control over '''exception handling''' and '''memory management''', and access to '''operating system facilities'''. The kernel can be divided into 5 logical clusters of classes:
* The first cluster contains the universal class defining facilities accessible to all other classes: [[ref:libraries/base/reference/any_chart|ANY]] . Every developer-defined class is a descendant of this class.
* The second cluster includes classes whose facilities are directly related to language concepts:
# The first cluster contains the universal class defining facilities accessible to all other classes: [[ref:libraries/base/reference/any_chart|ANY]] . Every developer-defined class is a descendant of this class.
# The second cluster includes classes whose facilities are directly related to language concepts:
** Classes describing the basic types: [[ref:libraries/base/reference/boolean_chart|BOOLEAN]] , [[ref:libraries/base/reference/character_8_chart|CHARACTER]] , [[ref:libraries/base/reference/integer_32_chart|INTEGER]] , [[ref:libraries/base/reference/real_32_chart|REAL]] and [[ref:libraries/base/reference/real_64_chart|DOUBLE]]
** Arrays: class [[ref:libraries/base/reference/array_chart|ARRAY]]
** Tuples: class [[ref:libraries/base/reference/tuple_chart|TUPLE]]
** Strings: class [[ref:libraries/base/reference/string_8_chart|STRING]]
** Basic facilities: class [[ref:libraries/base/reference/basic_routines_chart|BASIC_ROUTINES]]
* The third cluster provides input and output facilities:
# The third cluster provides input and output facilities:
** [[ref:libraries/base/reference/std_files_chart|STD_FILES]] offers basic mechanisms, sufficient for simple input and output.
** [[ref:libraries/base/reference/file_chart|FILE]] describes the notion of sequential file, viewed as a sequence of characters and fully integrated in the data structure library.
** [[ref:libraries/base/reference/directory_chart|DIRECTORY]] gives properties of directories (files serving as collections of other files).
* The next cluster, through class [[ref:libraries/base/reference/storable_chart|STORABLE]] , makes it possible to store object structures on persistent storage and retrieve them later. This facility can also be used to transmit object structures through pipes or over a network.
* The last cluster provides access to internal properties of the compiler and environment, useful for applications that need some fine-tuning of the basic mechanisms:
# The next cluster, through class [[ref:libraries/base/reference/storable_chart|STORABLE]] , makes it possible to store object structures on persistent storage and retrieve them later. This facility can also be used to transmit object structures through pipes or over a network.
# The last cluster provides access to internal properties of the compiler and environment, useful for applications that need some fine-tuning of the basic mechanisms:
** Class [[ref:libraries/base/reference/exceptions_chart|EXCEPTIONS]] (complemented by [[ref:libraries/base/reference/unix_signals_chart|UNIX_SIGNALS]] for Unix-type platforms) provides control over the exception handling mechanism, in particular for applications that need to handle different types of exception in different ways.
** Similarly, classes [[ref:libraries/base/reference/memory_chart|MEMORY]] and [[ref:libraries/base/reference/gc_info_chart|GC_INFO]] provide ways to control the garbage collector and tailor it to specific needs.
** Class ARGUMENTS gives access to the command-line arguments.
=Universal Class and its Features=
The Eiffel inheritance mechanism is set up in such a way that every class is a descendant of a Kernel Library class called [[ref:libraries/base/reference/any_chart|ANY]] . The features of this class provide a number of generally applicable facilities covering such needs as comparison, copying and rudimentary input and output.
==The structure of universal classes==
Every class which has no inheritance clause is understood to have an inheritance clause of the form
<code>
inherit
ANY
</code>
As a result, every developer-defined class is a descendant of [[ref:libraries/base/reference/any_chart|ANY]] . You may introduce your own project specific features in [[ref:libraries/base/reference/any_chart|ANY]] so that all the classes of your system will be able to use these features.
==Using the universal classes==
If you need to rename or redefine a feature inherited from one of the universal classes, you should include an explicit inheritance clause, as in
<code>
class
C
inherit
ANY
rename
out as basic_out
redefine
print
end
...
feature
...
end
</code>
The features of [[ref:libraries/base/reference/any_chart|ANY]] are usable in both qualified and unqualified form. For example, the argumentless function out, which produces a printable representation of any object, may be called under either of the forms
<code>
x := out
x := a.out
</code>
The first call yields a printable representation of the current object; the second, which assumes that a is not void, yields a printable representation of the object attached to a.
==Input and output features==
Some of the features of [[ref:libraries/base/reference/any_chart|ANY]] cover common input and output needs.
Feature <eiffel>io</eiffel>, of type [[ref:libraries/base/reference/std_files_chart|STD_FILES]] , gives access to standard input and output facilities. For example, <eiffel>io.input</eiffel> is the standard input file and <eiffel>io</eiffel>.<eiffel>new_line</eiffel> will print a line feed on the standard output. Feature <eiffel>io</eiffel> is declared as a once function which, when first called, returns the value of an instance of [[ref:libraries/base/reference/std_files_chart|STD_FILES]] that provides access to the standard input, the standard output and the error output. As a result, <eiffel>io</eiffel> is never void, so that operations such as <eiffel>io</eiffel>.<eiffel>new_line</eiffel> are always possible.
Function <eiffel>out</eiffel>, of type [[ref:libraries/base/reference/string_8_chart|STRING]] , is a universal mechanism for obtaining a simple external representation of any object. For non-void <code>x</code> of any type, the string <code>x.out</code> is a printable representation of <code>x</code>. This works for <code>x</code> of all types, reference or expanded. For example, if <code>x</code> is an integer expression, <code>x.out</code> is its string representation, such as <code>-897</code>; if n is a non-void reference, <code>x.out</code> is (recursively) the concatenation of the result of applying out to the successive fields of the attached object, each labeled by the name of the corresponding attribute. You may redefine out in any class to specify any suitable format for displaying instances of the class. To obtain the default representation regardless of any redefinition of out, use tagged_out, declared as a frozen synonym of the original out.
The call <code>print (x)</code> will output the value of <code>x.out</code> on the default output if <code>x</code> is not void, and do nothing otherwise.
==Copy and comparison routines==
Procedure copy copies the fields of an object onto those of another. It is used under the form
<code>
target.copy (source)
</code>
Here both target and source must be non-void; this means that <eiffel>copy</eiffel> is only good for copying onto an object that already exists. If you need both to allocate a new object and to initialize it as a copy of another, use the function <code>twin</code>. For non-void source, the assignment
<code>
target := source.twin
</code>
starts by creating a new object, then populates its fields to be identical to <code>source</code>.
The boolean function <eiffel>equal</eiffel> compares two objects for field-by-field equality. This is different from the equality operators <code>=</code> and <code>/=</code> which, in the case of reference types, compare references, not objects.
The function <eiffel>deep_twin</eiffel> produces a duplicate of an entire object structure. The boolean function <eiffel>deep_equal</eiffel> determines whether two object structures are recursively identical. These routines are the ''deep'' counterparts of the shallow copy and equality tests provided by <eiffel>twin</eiffel> and <eiffel>equal</eiffel>.
A class that needs a specific notion of equality and the corresponding copy semantics may redefine <eiffel>copy</eiffel> and <eiffel>is_equal</eiffel> (from which <code>equal</code> follows, since <code>equal (a, b)</code> is defined as <code>a.is_equal (b)</code> for non-void <code>a</code>). You will find such redefinitions in a number of classes of the Base libraries. For example an instance of [[ref:libraries/base/reference/string_8_chart|STRING]] is a string descriptor containing a reference to the actual character sequence, not that sequence itself, so that what the default equal compares and the default copy copies is the descriptor, not the string. Class [[ref:libraries/base/reference/string_8_chart|STRING]] redefines these routines to yield the semantics normally expected by string clients; the frozen variants <eiffel>standard_copy</eiffel> and <eiffel>standard_equal</eiffel>, originally declared as synonyms to <eiffel>equal</eiffel> and <eiffel>copy</eiffel>, remain available with the default semantics.
The function <eiffel>twin</eiffel> is defined in terms of <eiffel>copy</eiffel>, and so will follow any redefinition of <eiffel>copy</eiffel>. This makes it impossible to change the semantics of one but not of the other, which would be a mistake. The variant <eiffel>standard_twin</eiffel> is defined in terms of <eiffel>standard_copy</eiffel>.
{{note|In some existing Eiffel code you may encounter the use of a function <code>clone</code> which is used to do the job of <code>twin</code>, but has a form like <code>copy</code>, as in <code>target.clone (source)</code>. <code>clone</code> is an obsolete function. Use <code>twin</code> instead.
}}
==Type information==
The string-valued query <eiffel>generator</eiffel>, applied to any object, returns the name of the object's generating class: the class of which it is an instance. The boolean function <eiffel>conforms_to</eiffel> makes it possible to test dynamically whether the type of an object conforms to that of another - that is to say whether the first one's generator is a descendant of the second one's.
These two features enable clients to ascertain the dynamic type of an entity at runtime. They are only useful for low-level components; the normal mechanism for type-dependent operations is dynamic binding.
==Miscellaneous==
The query Void, of type <eiffel>NONE</eiffel>, denotes a reference that is always void - not attached to any object.
Procedure <eiffel>do_nothing</eiffel> does what its name implies.
Function default also has an empty body; its result type is <code>like Current</code>, so what it returns is the default value of the current type. This is mostly interesting for expanded types, since for reference types the default value is simply a void reference.
=Language-related Facilities=
A number of classes offer facilities which are very close to the language level. Here too the book ''[[Eiffel: The Language]]'' covers the classes in detail, so we can satisfy ourselves with a quick summary; the flat-short forms appear in part C.
==Basic types==
The basic types [[ref:libraries/base/reference/boolean_chart|BOOLEAN]] , [[ref:libraries/base/reference/character_8_chart|CHARACTER]] , [[ref:libraries/base/reference/integer_32_chart|INTEGER]] , [[ref:libraries/base/reference/real_32_chart|REAL]] and [[ref:libraries/base/reference/real_64_chart|DOUBLE]] are defined by classes of the Kernel library.
In reading the class specifications for the numeric types [[ref:libraries/base/reference/integer_32_chart|INTEGER]] , [[ref:libraries/base/reference/real_32_chart|REAL]] and [[ref:libraries/base/reference/real_64_chart|DOUBLE]] , you might think that the type declarations are too restrictive. For example the addition operation in class [[ref:libraries/base/reference/real_32_chart|REAL]] reads
<code>
infix "+" (other: REAL): REAL
</code>
but there is actually no problem here. A language convention applicable to all arithmetic expressions, the Balancing rule, states that in any such expression all operands are considered to be converted to the heaviest type, where [[ref:libraries/base/reference/real_64_chart|DOUBLE]] is heavier than [[ref:libraries/base/reference/real_32_chart|REAL]] and [[ref:libraries/base/reference/real_32_chart|REAL]] is heavier than [[ref:libraries/base/reference/integer_32_chart|INTEGER]] . So mixed-type arithmetic, consistent with common practice, is possible and indeed frequent.
==Arrays==
To create and manipulate one-dimensional arrays, use class [[ref:libraries/base/reference/array_chart|ARRAY]] of the Kernel Library. Arrays are not primitive language elements; instead, they are handled through class [[ref:libraries/base/reference/array_chart|ARRAY]] . This class is 'normal' in the sense that it may be used just as any other class by client and descendant classes. It is also somewhat special, however, in that the Eiffel compiler knows about it and uses this knowledge to generate efficient code for array operations.
To create an instance of [[ref:libraries/base/reference/array_chart|ARRAY]] , use the creation instruction
<code>
create my_array.make (1, u)
</code>
where the arguments indicate the lower and upper bounds. These bounds will then be accessible as <code>my_array</code> <code>.</code>lower and <code>my_array</code> <code>.</code><eiffel>upper</eiffel>. The number of items is <code>my_array</code> <code>.</code><eiffel>count</eiffel>; feature <eiffel>capacity</eiffel> is a synonym for <eiffel>count</eiffel>. The class invariant expresses the relation between <eiffel>count</eiffel>, <eiffel>lower</eiffel> and <eiffel>upper</eiffel>.
To access and change the item at index ''i'' in array a, you may use features <eiffel>item</eiffel> and <eiffel>put</eiffel>, as in
<code>
x := my_array.item (i)
my_array.put (new_value, i)
</code>
Function item has a bracket alias, so that you may also write the first assignment above more concisely as
<code>
x := my_array [i]
</code>
Features <eiffel>item</eiffel> and <eiffel>put</eiffel> have preconditions requiring the index ( <code>i</code>in the above calls) to be within the bounds of the array. This means that you can detect bounds violations (which correspond to bugs in the client software) by using a version of class [[ref:libraries/base/reference/array_chart|ARRAY]] compiled with precondition checking on. The bounds of an array may be changed dynamically through procedure <eiffel>resize</eiffel>. Previously entered elements are retained. Rather than an explicit resize, you may use calls to procedure <eiffel>force</eiffel> which has the same signature as put but no precondition; if the index is not within the current bounds force will perform a resize as necessary.
==Optimizing array computations==
''' CAUTION''': Although [[ref:libraries/base/reference/array_chart|ARRAY]] benefits from an efficient implementation, its more advanced facilities such as resizing do not come for free. For extensive computations on large arrays, an optimization may be desirable, bypassing these facilities. The technique yields loops that run at about the same speed as the corresponding loops written in C or Fortran (the usual references for array computations). It is of interest for advanced uses only, so that you may safely skip this section on first reading unless your domain of application is numerical computation or some other area requiring high-performance array manipulations.
The optimization relies on the class SPECIAL, used internally by [[ref:libraries/base/reference/array_chart|ARRAY]] but of no direct interest to client developers in most common uses. With the declarations
<code>
my_array: ARRAY [SOME_TYPE]
direct_access: SPECIAL [SOME_TYPE]
</code>
you may use <eiffel>direct_access</eiffel> in lieu of 'my_array' within a critical loop, provided none of the operations may resize the array. Typically, the operations should only include put and item. In such a case you can use the following scheme:
<code>
direct_access:= my_array.area
-- The critical loop:
from
some_initialization
index := some_initial_index
until
index = some_final_index
loop
...
x := direct_access.item (index)
...
direct_access.put (some_value, index)
...
end
</code>
This replaces an original loop where the operations were on <code>my_array</code>. Feature <eiffel>area</eiffel> of [[ref:libraries/base/reference/array_chart|ARRAY]] gives direct access to the special object, an instance of SPECIAL, containing the array values. Features <eiffel>put</eiffel> and <eiffel>item</eiffel> are available in SPECIAL as in [[ref:libraries/base/reference/array_chart|ARRAY]] , but without the preconditions; in other words, you will not get any bounds checking. Instances of SPECIAL are always indexed from zero, in contrast with arrays, whose lower bound is arbitrary, 1 being the most common value. But rather than performing index translations (that is to say, subtracting <code>my_array</code> <code>.</code><eiffel>lower</eiffel> from index throughout the loop) it is preferable to use the following simple technique: if the lower bound 'lb' of <code>my_array</code> is 1 or another small integer, use 0 as lower bound instead when creating <code>my_array</code>, but only use the positions starting at 'lb'. You will waste a few memory positions (0 to lb-1), but will not have to change anything in your algorithm and will avoid costly subtractions.
It is important to note that this optimization, if at all necessary, should at most affect a few loops in a large system. You should always begin by writing your software using the normal [[ref:libraries/base/reference/array_chart|ARRAY]] facilities; then once you have the certainty that the software is correct, if you detect that a large array computation is hampering the efficiency of the system, you may apply the above technique to get the fastest performance out of that computation. The change to the software will be minimal - a few lines - and will be easy to undo if necessary.
==Tuples==
A new Kernel Library class is introduced: [[ref:libraries/base/reference/tuple_chart|TUPLE]] .
Alone among all classes, class TUPLE has a variable number of generic parameters. <code>TUPLE, TUPLE [X], TUPLE [X, Y], TUPLE [X, Y, Z]</code> and so on are all valid types, assuming valid types <code>X, Y, Z</code> and so on.
Conformance rules:
<code>[CONF1]
For n >= 0
TUPLE [U1, U2, ..., Un, Un+1] conforms to
TUPLE [U1, U2, ..., Un]</code>
(and hence to <code>TUPLE [T1, T2, ..., Tn]</code> if each of the Ui conforms to each of the Ti for <code>1 <= i <= n</code>.)
In particular all tuple types conform to [[ref:libraries/base/reference/tuple_chart|TUPLE]] , with no parameter.
<code>[CONF2]
For n >= 0 If *every* one of the types T1, T2, ..., Tn conforms
to a type T, then TUPLE [T1, T2, ..., Tn] conforms
to ARRAY [T]</code>
{{Definition|Tuple Type|A "tuple type" is any type based on class [[ref:libraries/base/reference/tuple_chart|TUPLE]] , i.e. any type of the form <code>TUPLE [T1, T2, ..., Tn]</code> for any n (including 0, for which there is no generic parameter). }}
{{note|CONF1 should be understood in terms of the underlying mathematical model. <br/> Mathematically, <code>TUPLE [T1, T2, ..., Tn]</code> is the set TUPLE n of all partial functions f from N+ (the set of non-negative integers) to <code>T1 U T2 U ... Tn</code>, such that:<br/>The domain of f contains the interval 1..n (in other words, f is defined for any i such that 1 <nowiki><</nowiki>&#61; i <nowiki><</nowiki>&#61; n). <br/>For 1 <nowiki><</nowiki>&#61; i <nowiki><</nowiki>&#61; n, f (i) is a member of Ti. }}
With this definition, TUPLE <code>n</code> is indeed a subset of TUPLE <code>n+1</code>, and in particular TUPLE <code>0</code>, the empty set, is a subset of TUPLE <code>n</code> for any n.)
Semantics: an instance of <code>TUPLE [T1, T2, ..., Tn]</code> is a tuple whose first element is an instance of <code>T1</code>, the second element being an instance of <code>T2</code> etc. (The precise definition is the mathematical one given in note 1.) Note that there can be more than n elements to the tuple: for example a tuple with first element 5 and second element "FOO" is an instance of all of the following tuple types: <code>TUPLE; TUPLE [INTEGER]; TUPLE [INTEGER, STRING]</code>.
It may seem restrictive at first to permit only one class, [[ref:libraries/base/reference/tuple_chart|TUPLE]] , to have an arbitrary number of actual generic parameters. Why not have a general mechanism for declaring any class <code>C</code> so that all of <code>C [X], C [X, Y]</code> etc. are valid? But in fact this is not really a restriction. To obtain this effect without any complicated language convention, just declare <code>C</code> as
<code>C [G -> TUPLE]</code> and then use the generic derivations
<code>C [TUPLE [X]]
C [TUPLE [X, Y]]
</code>
and so on. This also makes it possible to have the effect of some fixed parameters and some variable ones, as in <code>C [G, H, I -> TUPLE]</code> so we have all the necessary flexibility.)
==Tuple expressions==
Let e1, e2, ..., en be expressions of respective types T1, T2, ..., Tn. Then the expression <code>[e1, e2, ..., en]</code> denotes an instance of <code>TUPLE [T1, T2, ..., Tn]</code>, whose first element is e1, the second element being e2, etc.
Tuple expressions can be nested: whereas <code>[1, 2, 3]</code> is a tuple with three elements (representing an instance of <code>TUPLE [INTEGER, INTEGER, INTEGER]), [1, [2, 3]]</code> is a tuple with two elements, the second one itself a tuple; the overall expression represents an instance of <code>TUPLE [INTEGER, TUPLE [INTEGER, INTEGER]]</code>.
As a special case of tuple expression syntax, the delimiters <code>[</code> and <code>]</code> are replaced by parentheses for the tuple representing the actual argument list of a routine call (see section 4).
==Tuple features==
The exact specification of class [[ref:libraries/base/reference/tuple_chart|TUPLE]] will be described in an addition to ELKS. The principal features are:
* [[ref:libraries/base/reference/tuple_chart|count]] (number of significant elements)
* [[ref:libraries/base/reference/tuple_chart|item]] (i), with the obvious precondition: the i-th element, of type [[ref:libraries/base/reference/any_chart|ANY]] (since the value of i is not known at compile time); also first, second, third, fourth and fifth, of the appropriate types.
* [[ref:libraries/base/reference/tuple_chart|put]] (x, i), with the obvious precondition: replace i-th element with x. If argument x is not of the appropriate type T <code>i</code> there is no effect.
* [[ref:libraries/base/reference/tuple_chart|is_equal]] : redefined to consider only the first n elements, where n is the smaller length.
Other features under consideration include:
* stripped (i): a tuple of type <code>TUPLE [T1, T2, Ti-1, Ti+1, ..., Tn]</code>, derived from the current one by removing the i-th component, again with the obvious precondition.
* wedged (x, i): a tuple with one more element, inserted at position i.
* '''infix''' "+": tuple concatenation
* '''infix''' "++": element concatenation; t ++ x is the same thing as t.wedged (x, t.count + 1).
==What have we gained?==
First we have solved the only case in the Eiffel programming language in which an expression has no precisely defined type: polymorphic manifest arrays. We don't have manifest arrays any more, but manifest tuples, with a precisely defined type. No incompatibility is introduced thanks to rule CONF2. The original syntax for manifest arrays, <code>Result := <<e1, e2, ..., en>></code>, will continue to be supported.
Second, we can define functions that return multiple results. This is a quite significant increase in expressive power. No common language has that. (You have to go to Lisp and functional languages.) Just define <code>TUPLE [...]</code> as the result type; in the function, you will write things like
<code>
Result := [e1, e2, ..., en]
</code>
Also, from a theoretical viewpoint, feature calls are simpler and more homogeneous: every feature takes exactly one tuple as argument and returns exactly one tuple as a result. (Either of these tuples may be empty: the first for a feature with no argument, the second for a procedure.) The syntax for a call becomes
<code>Feature Arguments</code>
with Arguments defined as
<code>Tuple_expression</code>
where the Tuple_expression uses the form given in section 2 but with the outermost <code>[</code> and <code>]</code> delimiters replaced by parentheses to conform to usual practice. So the call
<code>f (a, b, c)</code>
which we continue to think of as having three arguments a, b and c, formally has only one tuple argument <code>[a, b, c]</code>. This is of course not to be confused with a call of the form
<code>g ([a, b, c])</code>
which has one argument (a tuple with three elements) in both the ordinary and the formal sense.
==Active, iterators, numerical applications, introspection==
For a set of important applications of tuples see the book chapter on [[EiffelBase, Iteration|agents and iterators]] which also covers aspects of numerical software and related topics following from the tuple mechanism.
==Temporary limitations==
The implementation of tuples has the following limitations:
* Conformance of [[ref:libraries/base/reference/array_chart|ARRAY]] types to [[ref:libraries/base/reference/tuple_chart|TUPLE]] types is not yet fully supported.
* Class [[ref:libraries/base/reference/tuple_chart|TUPLE]] does not have features such as first and second. You must use item and, in most cases, an object test.
==Strings==
Strings are handled by class [[ref:libraries/base/reference/string_8_chart|STRING]] , similar in many respects to [[ref:libraries/base/reference/array_chart|ARRAY]] . Strings are of arbitrary size. The make creation procedure takes an integer argument, as in:
<code>
s, s1, s2, s3: STRING
...
create s.make (30)</code>
The argument indicates the number of characters for the initial allocation. This is not an absolute limit: the string will automatically grow or shrink as a result of future operations. You may always request a resizing explicitly by calling procedure resize.
==String descriptor==
The object attached at run-time to an entity such declared of type [[ref:libraries/base/reference/string_8_chart|STRING]] is not the actual sequence of characters but a string descriptor, which contains a reference to the actual string contents.
As a result, four assignment or assignment-like operations are possible:
* '''A1''' <code>s1 := s</code>
* '''A2''' <code>s2.share (s)</code>
* '''A3''' <code>s3 := s.twin</code>
* '''A4''' <code>s4.copy (s)</code>
As illustrated below, ''' A1''' is a reference assignment: <code>s1</code> will be attached to the same descriptor as s. ''' A2''' keeps the descriptors distinct, but make them refer to the same sequence of characters. In ''' A3''', <code>s3</code> will be attached to a new string, completely distinct from the string attached to <code>s1</code> although made of identical characters. ''' A4''' has almost the same effect as '''A3''', but is only applicable if <code>s4</code> was not void, and will override the existing descriptor rather than creating a new one.
[[Image:strings]]
fig. 1: Effect of string assignment and copy operations
[[ref:libraries/base/reference/basic_routines_chart|BASIC_ROUTINES]] provides a number of conversion functions, such as charconv.
=Files, Input, Output=
A few classes of the Kernel Library support file manipulation, input and output: [[ref:libraries/base/reference/std_files_chart|STD_FILES]] , FILE, [[ref:libraries/base/reference/directory_chart|DIRECTORY]] and [[ref:libraries/base/reference/unix_file_info_chart|UNIX_FILE_INFO]] . For simple applications it suffices to use [[ref:libraries/base/reference/std_files_chart|STD_FILES]] , but to understand the concepts better it is preferable to look first at the other two.
==General files==
FILE describes the notion of sequential file viewed as a data structure which fits in the general taxonomy of EiffelBase.
The class declaration defines files as unbounded sequences of characters. This means that you will find in FILE all the operations on sequential data structures that you have come to know and love by reading this documentation - at least, all that apply. Just as stacks and linked lists, files have <eiffel>put</eiffel>, <eiffel>extend</eiffel>, <eiffel>has</eiffel>, <eiffel>item</eiffel> and so on. More specific to files are the typed input and output operations. For output, you will find <eiffel>put_character</eiffel>, <eiffel>put_integer</eiffel>, <eiffel>put_real</eiffel>, <eiffel>put_double</eiffel> and <eiffel>put_string</eiffel>, as well as <eiffel>new_line</eiffel>. For input you will find <eiffel>read_integer</eiffel> and its co-conspirators.
{{caution|Note the application to input features of the command-query separation principle. <br/>
The input features such as read_integer do not by themselves return a result; they set the values of queries such as last_integer. So the normal way to read is through two operations:<br/>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;my_file.read_integer<br/>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;new_value :&#61; my_file.last_integer}}
Queries are available to determine the status of a file, in particular <eiffel>exists</eiffel>, <eiffel>is_readable</eiffel>, <eiffel>is_executable</eiffel>, <eiffel>is_writable</eiffel>, <eiffel>is_creatable</eiffel>, <eiffel>is_closed</eiffel>, <eiffel>is_open_read</eiffel> and so on.
{{caution|You will notice in the flat-short form that all these queries except the first have exists as a precondition. This precondition is good for efficiency since it saves an existence test - a relatively expensive operation - when you know that a certain file exists. But it also means that if you have any doubt about the file's existence you must use the queries in the style<br/>
<eiffel>if my_file.exists and then my_file.is_readable then ...</eiffel>}}
FILE is a deferred class. Various implementations are possible. A quite detailed one is PLAIN_TEXT_FILE, which adds many features for accessing reading and writing data from/to a file.
[[ref:libraries/base/reference/unix_file_info_chart|UNIX_FILE_INFO]] describes objects that contain internal information, such as protection mode and size, about a file.
The class [[ref:libraries/base/reference/directory_chart|DIRECTORY]] describes those files which are directories - nodes in the tree describing the file structure.
==Basic input and output==
Regardless of the operating system that you use, for simple input and output [[ref:libraries/base/reference/std_files_chart|STD_FILES]] is sufficient. You may inherit from that class to gain direct access to its features; or you may declare an entity of type [[ref:libraries/base/reference/std_files_chart|STD_FILES]] . But remember that a feature of this type is always available: io, from class [[ref:libraries/base/reference/any_chart|ANY]] . Thanks to this feature you may include simple input and output in any class, with instructions such as
<code>io.put_string ("My message")</code>
[[ref:libraries/base/reference/std_files_chart|STD_FILES]] defines three default files through features <eiffel>input</eiffel>, <eiffel>output</eiffel> and <eiffel>error</eiffel>. These features are Once functions, so that the first reference to any one of them will automatically create the corresponding file descriptor and open the associated file.
To simplify the writing of common input and output operations, the most frequently used features of class FILE - for reading and writing integers, reals and so on, as discussed next - have been repeated in [[ref:libraries/base/reference/std_files_chart|STD_FILES]] so as to apply to the default input and output. Procedure put_string in the example at the beginning of this section is typical: it writes its output on the standard output. More generally, [[ref:libraries/base/reference/std_files_chart|STD_FILES]] has all the <eiffel>put_xxx</eiffel>, <eiffel>read_xxx</eiffel> and <eiffel>last_xxx</eiffel> features of FILE.
=PERSISTENCE, STORAGE, AND RETRIEVAL=
Most object-oriented applications need the ability to store object structures on persistent storage for later retrieval, and to transfer such object structures to other applications.
Class [[ref:libraries/base/reference/storable_chart|STORABLE]] addresses this need.
==Persistence completeness==
A fundamental requirement on object persistence mechanisms is the '' Persistence Completeness'' rule, stated as follows in ''[[Eiffel: The Language]]'':
Whenever a routine of class [[ref:libraries/base/reference/storable_chart|STORABLE]] stores an object into an external file, it stores with it the dependents of that object. Whenever one of the associated retrieval routines retrieves a previously stored object, it also retrieves all its dependents.
Storing an object just by itself would usually result in wrong semantics: most objects contain references to other objects, which must also be stored and retrieved with it. The persistence completeness rule ensures that this is always the case. It also means, of course, that the features of [[ref:libraries/base/reference/storable_chart|STORABLE]] must do much more than simple input and output; they must perform complete traversals of object structures.
==Using the storage and retrieval facilities==
Class [[ref:libraries/base/reference/storable_chart|STORABLE]] is meant to be used as ancestor. You can use its features in any descendant <eiffel>C</eiffel>; for example a routine of <eiffel>C</eiffel> may contain a call of the form <eiffel>basic_store (my_descriptor)</eiffel>.
The effect of this call will be to store the current object and all its dependents into the file denoted by my_descriptor.
Although <eiffel>basic_store</eiffel> and other procedures of [[ref:libraries/base/reference/storable_chart|STORABLE]] will in general process objects of many different types, only the generating class of the structure's initial object, <eiffel>C</eiffel> in our example, needs to be a descendant of [[ref:libraries/base/reference/storable_chart|STORABLE]] .
==Varieties of store operations==
Different variants of the store operation are supported: basic store and independent store. Basic store produces more compact structures in the resulting files, and is slightly faster; but the resulting structure is dependent on the system which executes the store operation ('System' is taken here, as elsewhere in this documentation, in its Eiffel sense of an executable assembly of classes, compiled together with the help of a configuration file.) This means that you can use procedure <eiffel>basic_store</eiffel> to store an object structure during an execution of a system if you will only retrieve it later in that execution, or in a subsequent execution of the same system. Another variant, <eiffel>independent_store</eiffel>, uses an entirely platform-independent storage format. It allows a system running on a computer with a certain architecture to retrieve, without any explicit conversion operation, object structures stored by a system running on a machine of a completely different architecture.
==Retrieval==
You only need to be aware of the difference between basic and general store at storage time. The stored structure will always be available through feature retrieved; this feature will figure out, from the format of the stored structure, whether it was stored by <eiffel>basic_store</eiffel>, or <eiffel>independent_store</eiffel> and will decode it accordingly.
Feature <eiffel>retrieved</eiffel> returns a result of type [[ref:libraries/base/reference/any_chart|ANY]] and is typically used through an object test of the form
<code>
if attached {MY_TYPE} retrieved (my_descriptor) as data then
-- Retrieved result is of expected type.
-- Proceed with processing of result,
-- typically with calls of the form `data.some_feature'.
else
-- Result was not of expected type MY_TYPE.
end
</code>
The object test is necessary because <eiffel>retrieved</eiffel> returns a result of type [[ref:libraries/base/reference/any_chart|ANY]] whereas the type of <code>x</code> will be based on a proper descendant of [[ref:libraries/base/reference/any_chart|ANY]] . If the structure in the file has been corrupted and <eiffel>retrieved</eiffel> is unable to do its job, it will trigger an exception. The code for that exception in class [[ref:libraries/base/reference/exceptions_chart|EXCEPTIONS]] (which inherits it from EXCEP_CONST and is discussed in the next section, together with the notion of exception code) is <eiffel>Retrieve_exception</eiffel>.
==Recoverable storable==
Sometimes you will be in a position where the schema of a class will have changed between the time you stored your object and the time you are trying to retrieve it. Such changes include:
* class name changed
* attributes have been added or removed
* attributes have been renamed
* attributes type have changed
The storable mechanism allows you to retrieve the old version of the object only if it was saved using the <eiffel>independent_store</eiffel> facility. Each time you retrieve an object of a certain base class whose schema has changed, the feature <eiffel>correct_mismatch</eiffel> will be called. This feature is defined in [[ref:libraries/base/reference/any_chart|ANY]] and by default will raise an exception. To handle the mismatch, you need to redefine <eiffel>correct_mismatch</eiffel> in the base class whose schema has been changed. For example in EiffelBase, [[ref:libraries/base/reference/hash_table_chart|HASH_TABLE]] has changed between version 5.1 and version 5.2 to use <eiffel> SPECIAL</eiffel> rather than [[ref:libraries/base/reference/array_chart|ARRAY]] for its internal data storage. To retrieve a 5.1 version of [[ref:libraries/base/reference/hash_table_chart|HASH_TABLE]], you need to define <eiffel>correct_mismatch</eiffel> as following:
<code>
correct_mismatch
-- Attempt to correct object mismatch during retrieve using `mismatch_information'.
do
-- In version 5.1 and earlier, `content', `keys' and `deleted_marks'
-- were of base class ARRAY. In 5.2 we changed it to be a SPECIAL for
-- efficiency reasons. In order to retrieve an old HASH_TABLE we
-- need to convert those ARRAY instances into SPECIAL instances.
-- Convert `content' from ARRAY to SPECIAL.
if attached {ARRAY [G]} mismatch_information.item ("content") as l_temp then
content := l_temp.area
end
-- Convert `keys' from ARRAY to SPECIAL.
if attached {ARRAY [H]} mismatch_information.item ("keys") as l_temp then
keys := l_temp.area
end
-- Convert `deleted_marks' from ARRAY to SPECIAL.
if attached {ARRAY [BOOLEAN]} mismatch_information.item ("deleted_marks") as l_temp then
deleted_marks := l_temp.area
end
if content = Void or keys = Void or deleted_marks = Void then
-- Could not retrieve old version of HASH_TABLE. We throw an exception.
Precursor {TABLE}
end
end
</code>
Note the use of <eiffel>mismatch_information</eiffel>, this is a once feature of [[ref:libraries/base/reference/any_chart|ANY]] of type <eiffel>MISMATCH_INFORMATION</eiffel> which behaves like a [[ref:libraries/base/reference/hash_table_chart|HASH_TABLE]]. The keys of the table are the names of the attributes on which a mismatch occurred and the values are the corresponding object fields as they were originally stored. In this particular case of [[ref:libraries/base/reference/hash_table_chart|HASH_TABLE]] we know that the previous version was an [[ref:libraries/base/reference/array_chart|ARRAY]], so we do an object test and if it succeeds we assign its <eiffel>area</eiffel> to the corresponding attribute of [[ref:libraries/base/reference/hash_table_chart|HASH_TABLE]].
If a class name changed, then you need to create an instance of <eiffel>CLASS_NAME_TRANSLATIONS</eiffel>, it behaves like a [[ref:libraries/base/reference/hash_table_chart|HASH_TABLE]] where the keys represent the old name, and the value the new name. This instance needs to be created before the call to retrieved.
=Access To Internal Properties=
In some applications, you may need to fine-tune the exception handling and memory management mechanisms. You may also need a simple way to access command-line arguments. In less common cases, you may require low-level access to internal properties of objects.
==Exception handling==
Class [[ref:libraries/base/reference/exceptions_chart|EXCEPTIONS]] enables you to control the handling of exceptions. [[ref:libraries/base/reference/unix_signals_chart|UNIX_SIGNALS]] , discussed next, complements it for the special case of fine-grain signal handling on Unix or Unix-like platforms. Both are meant to be inherited by any class that needs their facilities.
The basic exception mechanism treats all exceptions in the same way. In some cases, it may be useful to discriminate in a Rescue clause between the various possible causes.
Class [[ref:libraries/base/reference/exceptions_chart|EXCEPTIONS]] provides the features to do this. Each kind of exception has an integer code, which you can use through several features:
* The integer-valued query exception which gives the code of the latest exception.
* Queries which determine the general nature of the latest exception: <eiffel>is_signal</eiffel> which determines whether the exception was an operating system signal; <eiffel>is_developer_exception</eiffel> which determines whether it was explicitly caused by a raise, as explained next; <eiffel>assertion_violation</eiffel>.
* Query<eiffel> recipient_name</eiffel> which gives the name of the exception's recipient - the routine that was interrupted by the exception.
The class also provides a set of constant integer-valued attributes which denote the various possible codes, such as <eiffel>No_more_memory</eiffel>, <eiffel>Routine_ failure</eiffel> and <eiffel>Precondition_violation</eiffel>. So you can test the value of exception against these codes if you need to ascertain the precise nature of an exception. To keep [[ref:libraries/base/reference/exceptions_chart|EXCEPTIONS]] simple these constant attributes are declared in a class EXCEP_CONST, of which [[ref:libraries/base/reference/exceptions_chart|EXCEPTIONS]] is an heir.
Another occasional requirement is for a mechanism to trigger an exception explicitly. Procedure raise answers this needs; the argument, a string, is the tag chosen for the exception. The code in this case is <eiffel>Developer_exception</eiffel>; the query <eiffel>is_developer_exception</eiffel> will return true; and the tag is accessible through feature <eiffel>tag_name</eiffel>.
You will notice in the interface specification for [[ref:libraries/base/reference/exceptions_chart|EXCEPTIONS]] that for some properties of the latest exception there are two features, one with a name such as exception or recipient_name as seen above and the other with a name prefixed by original_: <eiffel>original_exception</eiffel>, <eiffel>original_recipient_name</eiffel>.
{{caution|The reason for the presence of these pairs is that the immediately visible cause of a routine interruption may not be the real one. Assume that routine <eiffel>r</eiffel> from class <eiffel>C</eiffel>, which has a Rescue clause, calls <eiffel>s</eiffel> from <eiffel>D</eiffel> with no Rescue clause, and that some call executed by <eiffel>s</eiffel> causes a precondition violation. Because <eiffel>s</eiffel> has no Rescue clause of its own, <eiffel>s</eiffel> will fail. Up the call chain, the first routine that has a Rescue clause - <eiffel>r</eiffel> itself, or one of its own direct or indirect callers - may process the exception; but if it examines the exception code through attribute exception it will get the value of <eiffel>Routine_failure</eiffel>. This may be what you want; but to handle the situation in a finer way you will usually need to examine the code for the original exception, the one that interrupted <eiffel>s</eiffel>. This code will be accessible through the attribute original_exception, which in this case will have the value of <eiffel>Precondition</eiffel>, the exception code for precondition violations. So you have the choice between exploring the properties of the original exception, or those of the resulting routine failures. Just make sure you know what you are looking for. }}
As you will see from the header comments in the flat-short form of class [[ref:libraries/base/reference/exceptions_chart|EXCEPTIONS]] , the queries that return detailed information about an exception, such as <eiffel>assertion_violation</eiffel>, all give an answer determined by <eiffel>original_exception</eiffel> rather than <eiffel>exception</eiffel>, since when the two are different (that is to say, when you handle the exception in a routine other than the original recipient) the value of exception is always <eiffel>Routine_failure</eiffel> and there is nothing more to say about it.
==Signal handling==
The features of class [[ref:libraries/base/reference/exceptions_chart|EXCEPTIONS]] enable you to determine whether a certain exception is a signal - an operating system event such as may result from a child process that disappears, a window that is resized, a user that hits the Break key and many others. But they do not give you more details because the exact set of possible signals is highly platform-dependent.
Class [[ref:libraries/base/reference/unix_signals_chart|UNIX_SIGNALS]] complements EXCEP_CONST by providing codes for the signals of Unix and similar systems, such as Sigkill for the 'kill' signal and Sigbus for bus error.
Query <eiffel>is_defined </eiffel> <code>(some_signal)</code>, where <code>some_signal</code> is an integer code, will determine whether some_signal is supported on the platform.
A class whose routines need to perform specific processing depending on the nature of signals received should inherit from [[ref:libraries/base/reference/unix_signals_chart|UNIX_SIGNALS]] , or a similar class for another platform.
Because signal codes are platform-dependent, the features of [[ref:libraries/base/reference/unix_signals_chart|UNIX_SIGNALS]] are implemented as once functions - computed on the first call - rather than constants, although this makes no difference to clients.
==Memory management==
Class [[ref:libraries/base/reference/memory_chart|MEMORY]] , like [[ref:libraries/base/reference/exceptions_chart|EXCEPTIONS]] , is meant to be used as an ancestor by classes that need its facilities. It offers a number of features for controlling memory management and fine-tuning the garbage collection mechanism, a key component of the Eiffel Software environment.
One of the most useful features in this class is <e>dispose</e>. This procedure describes actions to be applied to an unreachable object just before the garbage collector reclaims it. By default, as declared in [[ref:libraries/base/reference/memory_chart|MEMORY]] , the procedure does nothing; but you may redefine it in a proper descendant of [[ref:libraries/base/reference/memory_chart|MEMORY]] to describe dispose actions. Normally such actions will involve freeing external resources: for example a class describing file descriptors may redefine <e>dispose</e> so that whenever a descriptor object is garbage-collected the corresponding file will be closed.
{{caution|This example is typical of proper uses of <e>dispose</e>. In a dispose <e>procedure</e>, you should not include any instruction that could modify the Eiffel object structure, especially if some objects in that structure may themselves have become unreachable: these instructions could conflict with the garbage collector's operations and cause catastrophic behavior. The legitimate use of <e>dispose</e> redefinition is for disposing of non-Eiffel resources. }}
Other features of [[ref:libraries/base/reference/memory_chart|MEMORY]] provide direct control over the operation of the garbage collector. You can in particular stop garbage collection through a call to <eiffel>collection_off</eiffel>, and restart it through a call to <eiffel>collection_on</eiffel>. By default, garbage collection is always on (a testimony to its authors' trust in its efficiency). Garbage collection is normally incremental, so as not to disrupt the application in a perceptible way. To start a complete garbage collection mechanism - reclaiming all unused objects - call procedure <eiffel>full_collect</eiffel>. The remaining features of [[ref:libraries/base/reference/memory_chart|MEMORY]] enable finer control of the collection mechanism and are useful in special cases only. You will even find a free procedure providing brave (and competent) developers with a mechanism for reclaiming individual objects manually.
MEM_INFO, the result type for query <eiffel>memory_statistics</eiffel> in [[ref:libraries/base/reference/memory_chart|MEMORY]] , describes objects containing information collected about memory usage. The features of [[ref:libraries/base/reference/gc_info_chart|GC_INFO]] provide statistics about the garbage collector's operation.
==Command-line arguments==
Writing, assembling and compiling a system yields an executable command. The system's users will call that command with arguments. These are normally provided in textual form on the command line, as in
<code>your_system arg1 arg2 arg3</code>
although one may conceive of other ways of entering the command arguments, such as tabular or graphical form-filling. In any case, the software must be able to access the values passed as command arguments.
A language mechanism is available for that purpose: the Root Class rule indicates that the creation procedure of the root class may have a single argument (in the Eiffel sense of argument to a routine) of type <code>ARRAY [STRING]</code>. The corresponding array of strings will be initialized at the beginning of the system's execution with the values entered as arguments to that execution of the command.
Although this facility suffices in many cases, it is not always convenient if you suddenly need to access the command arguments in a class that is far away from the root. An alternative mechanism, class ARGUMENTS, is available. Once again, this is a class from which you should inherit if you need its facilities. It has just two exported features:
* <eiffel>argument_count</eiffel>, a non-negative integer, is the number of command arguments.
* <eiffel>argument</eiffel> <code>(</code>i<code>)</code>, a string, is the i-th command argument. Here <code>i</code> must be between 0 and <eiffel>argument_count</eiffel>; the convention is that for <code>i </code> <code>= 0</code> the result is the name of the command itself.