Fix various minor documentation bugs.

git-svn-id: svn+ssh://svn.gna.org/svn/gnustep/libs/base/trunk@14777 72102866-910b-0410-8b05-ffd578937521
This commit is contained in:
Richard Frith-Macdonald 2002-10-14 11:05:49 +00:00
parent d54488d74d
commit b75da9fd95
6 changed files with 516 additions and 433 deletions

View file

@ -189,7 +189,7 @@
* class of the object to be aClass etc. The allocated memory will
* be extraBytes larger than the space actually needed to hold the
* instance variables of the object.<br />
* This function is used by the [NSObject-allocWithZone:] mnethod.
* This function is used by the [NSObject+allocWithZone:] mnethod.
*/
GS_EXPORT NSObject *
NSAllocateObject(Class aClass, unsigned extraBytes, NSZone *zone);
@ -206,7 +206,8 @@ NSDeallocateObject(NSObject *anObject);
* memory allocated from zone. The allocated memory will be extraBytes
* longer than that necessary to actually store the instance variables
* of the copied object.<br />
* This is used by the [NSObject-copyWithZone:] method.
* This is used by the NSObject implementation of the
* [(NSCopying)-copyWithZone:] method.
*/
GS_EXPORT NSObject *
NSCopyObject(NSObject *anObject, unsigned extraBytes, NSZone *zone);

View file

@ -510,7 +510,7 @@ GSGetValue(NSObject *self, NSString *key, SEL sel,
* supplied).<br />
* Automatic conversion between NSNumber and C scalar types is performed.<br />
* If type is null and can't be determined from the selector, the
* [NSObject-handleTakevalue:forUnboundKey:] method is called to try
* [NSObject-handleTakeValue:forUnboundKey:] method is called to try
* to set a value.
*/
void

View file

@ -188,8 +188,8 @@ static NSMutableSet *textNodes = nil;
if (u == nil)
{
u = unit;
s = base;
}
s = base;
}
else if (u == nil)
{
@ -215,7 +215,7 @@ static NSMutableSet *textNodes = nil;
s = [globalRefs unitRef: r type: t unit: &u];
}
}
else
if (s == nil)
{
NSString *tmp = u;
@ -1448,12 +1448,12 @@ static NSMutableSet *textNodes = nil;
NSString *type = [prop objectForKey: @"type"];
NSString *r = [prop objectForKey: @"id"];
GSXMLNode *tmp = [node firstChild];
NSString *c = [prop objectForKey: @"class"];
NSString *s;
if ([type isEqual: @"method"] || [type isEqual: @"ivariable"])
{
s = [prop objectForKey: @"class"];
s = [self makeLink: r ofType: type inUnit: s isRef: YES];
s = [self makeLink: r ofType: type inUnit: c isRef: YES];
}
else
{
@ -1470,7 +1470,16 @@ static NSMutableSet *textNodes = nil;
}
if (s == nil)
{
NSLog(@"ref '%@' not found for %@", r, type);
if (c == nil)
{
NSLog(@"No unique ref to %@ '%@' not found in %@",
type, r, base);
}
else
{
NSLog(@"ref to the %@ version of %@ '%@' not found in %@",
c, type, r, base);
}
if (tmp == nil)
{
[buf appendString: r];

View file

@ -69,6 +69,11 @@ static BOOL snuggleStart(NSString *t)
* far too special purpose.</p>
* <unit />
* <p>Here is the afterword for the class.</p>
* <p> Ad here is some automated cross referencing ...
* A method in a protocol: [(NSCopying)-copyWithZone:], a class:
* [NSString], a protocol: [(NSCopying)], and a
* category: [NSRunLoop(GNUstepExtensions)].
* </p>
* </unit>
* And finally, here is the actual class description ... outside the chapter.
* With a reference to foo() in it.
@ -1809,6 +1814,7 @@ static BOOL snuggleStart(NSString *t)
/*
* Ensure that methods are rendered as references.
* First look for format with class name in square brackets.
* If that's all there is, we make a class reference.
*/
r = [tmp rangeOfString: @"["];
if (r.length > 0)
@ -1824,6 +1830,7 @@ static BOOL snuggleStart(NSString *t)
NSString *cName = nil;
NSString *mName = nil;
unichar c;
BOOL isProtocol = NO;
if (pos < ePos
&& [identStart characterIsMember:
@ -1880,8 +1887,7 @@ static BOOL snuggleStart(NSString *t)
cName = [tmp substringWithRange: r];
}
}
else if (pos < ePos
&& (c = [tmp characterAtIndex: pos]) == '(')
else if (pos < ePos && (c = [tmp characterAtIndex: pos]) == '(')
{
/*
* Look for protocol name.
@ -1903,6 +1909,7 @@ static BOOL snuggleStart(NSString *t)
}
pos++;
}
isProtocol = YES;
}
if (pos < ePos && (c == '+' || c == '-'))
@ -1986,6 +1993,23 @@ static BOOL snuggleStart(NSString *t)
ref = [NSString stringWithFormat:
@"<ref type=\"method\" id=\"%@\">", mName];
}
else if (isProtocol == YES)
{
ref = [NSString stringWithFormat:
@"<ref type=\"method\" id=\"%@\" class=\"%@\">",
mName, cName];
if (isProtocol == YES)
{
if (sub == nil)
{
sub = tmp;
}
sub = [sub stringByReplacingString: @"("
withString: @"&lt;"];
sub = [sub stringByReplacingString: @")"
withString: @"&gt;"];
}
}
else
{
ref = [NSString stringWithFormat:
@ -2005,6 +2029,34 @@ static BOOL snuggleStart(NSString *t)
[a insertObject: end atIndex: ++l];
}
}
else if (pos == ePos && cName != nil)
{
NSString *ref;
if (isProtocol == YES)
{
NSRange r;
r = NSMakeRange(1, [cName length] - 2);
cName = [cName substringWithRange: r];
ref = [NSString stringWithFormat:
@"<ref type=\"protocol\" id=\"(%@)\">&lt;%@&gt;</ref>",
cName, cName];
}
else if ([cName hasSuffix: @")"] == YES)
{
ref = [NSString stringWithFormat:
@"<ref type=\"category\" id=\"%@\">%@</ref>",
cName, cName];
}
else
{
ref = [NSString stringWithFormat:
@"<ref type=\"class\" id=\"%@\">%@</ref>",
cName, cName];
}
[a replaceObjectAtIndex: l withObject: ref];
}
}
continue;
}

View file

@ -20,428 +20,449 @@
<chapter>
<heading>The autogsdoc tool</heading>
<p>
The autogsdoc tool is a command-line utility for parsing ObjectiveC
source code (header files and optionally source files) in order to
generate documentation covering the public interface of the various
classes, categories, and protocols in the source.
</p>
<p>
The simple way to use this is to run the command with one or more
header file names as arguments ... the tool will automatically
parse corresponding source files in the same directory as the
headers (or the current directory, or the directory specified
using the DocumentationDirectory default), and produce gsdoc
and html files as output.
</p>
<p>
Even without any human assistance, this tool will produce skeleton
documents listing the methods in the classes found in the source
files, but more importantly it can take specially formatted comments
from the source files and insert those comments into the gsdoc output.
</p>
<p>
Any comment beginning with slash and <em>two</em> asterisks rather than
the common slash and single asterisk, is taken to be gsdoc markup, to
be use as the description of the class or method following it. This
comment text is reformatted and then inserted into the output.<br />
Where multiple comments are associatd with the same item, they are
joined together with a line break (&lt;br /&gt;) between each if
necessary.
</p>
<p>
The tool can easily be used to document programs as well as libraries,
simply by giving it the name of the source file containing the main()
function of the program - it takes the special comments from that
function and handles them specially, inserting them as a section at
the end of the first chapter of the document (it creates the first
chapter if necessary).
</p>
<p>
There are some cases where special extra processing is performed,
predominantly in the first comment found in the source file,
from which various chunks of gsdoc markup may be extracted and
placed into appropriate locations in the output document -
</p>
<list>
<item><strong>AutogsdocSource</strong>
In any line where <code>AutogsdocSource</code>: is found, the remainder
of the line is taken as a source file name to be used instead of
making the assumption that each .h file processed uses a .m file
of the same name. You may supply multiple <code>AutogsdocSource</code>:
lines where a header file declares items which are defined in
multiple source files.<br />
If a file name is absolute, it is used just as supplied.<br />
If on the other hand, it is a relative path, the software looks for
the source file first relative to the location of the header file,
and if not found there, relative to the current directory in which
autogsdoc is running, and finally relative to the directory
specified by the DocumentationDirectory default.
</item>
<item><strong>&lt;abstract&gt;</strong>
An abstract of the content of the document ... placed in the head
of the gsdoc output.
</item>
<item><strong>&lt;author&gt;</strong>
A description of the author of the code - may be repeated to handle
the case where a document has multiple authors. Placed in the
head of the gsdoc output.<br />
As an aid to readability of the source, some special additional
processing is performed related to the document author -<br />
Any line of the form '<code>Author</code>: name &lt;email-address&gt;',
or '<code>By</code>: name &lt;email-address&gt;',
or '<code>Author</code>: name' or '<code>By</code>: name'
will be recognised and converted to an <em>author</em> element,
possibly containing an <em>email</em> element.
</item>
<item><strong>&lt;back&gt;</strong>
Placed in the gsdoc output just before the end of the body of the
document - intended to be used for appendices, index etc.
</item>
<item><strong>&lt;chapter&gt;</strong>
Placed immediately before any generated class documentation ...
intended to be used to provide overall description of how the
code being documented works.<br />Any documentation for the main()
function of a program is inserted as a section at the end of this
chapter.
</item>
<item><strong>&lt;copy&gt;</strong>
Copyright of the content of the document ... placed in the head
of the gsdoc output.<br />
As an aid to readability of the source, some special additional
processing is performed -<br />
Any line of the form 'Copyright (C) text' will be recognised and
converted to a <em>copy</em> element.
</item>
<item><strong>&lt;date&gt;</strong>
Date of the revision of the document ... placed in the head
of the gsdoc output. If this is omitted the tool will try to
construct a value from the RCS Date tag (if available).
</item>
<item><strong>&lt;front&gt;</strong>
Inserted into the document at the start of the body ... intended
to provide for introduction or contents pages etc.
</item>
<item><strong>&lt;title&gt;</strong>
Title of the document ... placed in the head of the gsdoc output.
If this is omitted the tool will generate a (probably poor)
title of its own - so you should include this markup manually.
</item>
<item>
<strong>NB</strong>This markup may be used within
class, category, or protocol documentation ... if so, it is
extracted and wrapped round the rest of the documentation for
the class as the classes chapter.
The rest of the class documentation is normally
inserted at the end of the chapter, but may instead be substituted
in in place of the &lt;unit /&gt; pseudo-element within the
&lt;chapter&gt; element.
</item>
<item><strong>&lt;version&gt;</strong>
Version identifier of the document ... placed in the head
of the gsdoc output. If this is omitted the tool will try to
construct a value from the RCS Revision tag (if available).
</item>
</list>
<p>
In comments being used to provide text for a method description, the
following markup is removed from the text and handled specially -
</p>
<list>
<item><strong>&lt;init /&gt;</strong>
The method is marked as being the designated initialiser for the class.
</item>
<item><strong>&lt;override-subclass /&gt;</strong>
The method is marked as being one which subclasses must override
(eg an abstract method).
</item>
<item><strong>&lt;override-never /&gt;</strong>
The method is marked as being one which subclasses should <em>NOT</em>
override.
</item>
<item><strong>&lt;standards&gt; ... &lt;/standards&gt;</strong>
The markup is removed from the description and placed <em>after</em>
it in the gsdoc output - so that the method is described as
conforming (or not conforming) to the specified standards.
</item>
</list>
<p>
Generally, the text in comments is reformatted to standardise and
indent it nicely ... the reformatting is <em>not</em> performed on
any text inside an &lt;example&gt; element.<br />
When the text is reformatted, it is broken into whitespace separated
'words' which are then subjected to some extra processing ...
</p>
<list>
<item>Certain well known constants such as YES, NO, and nil are
enclosed in &lt;code&gt; ... &lt;/code&gt; markup.
</item>
<item>The names of method arguments within method descriptions are
enclosed in &lt;var&gt; ... &lt;/var&gt; markup.
</item>
<item>Method names (beginning with a plus or minus) are enclosed
in &lt;ref...&gt; ... &lt;/ref&gt; markup.<br />
eg. "-init" (without the quotes) would be wrapped in a gsdoc
reference element to point to the init method of the current
class or, if only one known class had an init method, it
would refer to the method of that class.
<br />Note the fact that the method name must be surrounded by
whitespace (though a comma, fullstop, or semicolon at the end
of the specifier will also act as a whitespace terminator).
</item>
<item>Method specifiers including class names (beginning and ending with
square brackets) are enclosed in &lt;ref...&gt; ... &lt;/ref&gt; markup.
<br />eg. <code>[</code>NSObject-init<code>]</code>,
will create a reference to the init method of NSObject (either the
class proper, or any of its categories), while
<br /><code>[</code>(NSCopying)-copyWithZone:<code>]</code>, creates a
reference to a method in the NSCopyIng protocol.
<br />Note that no spaces must appear between the square brackets
in these specifiers.
<br />Protocol names are enclosed in round brackets rather than
the customary angle brackets, because gsdoc is an XML language, and
XML treats angle brackets specially.
</item>
<item>Function names (ending with '()') other than 'main()' are enclosed
in &lt;ref...&gt; ... &lt;/ref&gt; markup.<br />
eg. "NSLogv()" (without the quotes) would be wrapped in a gsdoc
reference element to point to the documentation of the NSLog function.
<br />Note the fact that the function name must be surrounded by
whitespace (though a comma, fullstop, or semicolon at the end
of the specifier will also act as a whitespace terminator).
</item>
</list>
<p>
The tools accepts certain user defaults (which can of course be
supplied as command-line arguments as usual) -
</p>
<list>
<item><strong>Clean</strong>
If this boolean value is set to YES, then rather than generating
documentation, the tool removes all gsdoc files generated in the
project, and all html files generated from them (as well as any
which would be generated from gsdoc files listed explicitly),
and finally removes the project index file.<br />
The only exception to this is that template gsdoc files (ie those
specifield using ConstantsTemplate, FunctionsTemplate arguments etc)
are not deleted unless the CleanTemplates flag is set.
</item>
<item><strong>CleanTemplates</strong>
This flag specifies whether template gsdoc files are to be removed
along with other files when the Clean option is specified.
The default is for them not to be removed ... since these templates
may have been produced manually and just had data inserted into them.
</item>
<item><strong>ConstantsTemplate</strong>
Specify the name of a template document into which documentation
about constants should be inserted from all files in the project.<br />
This is useful if constants in the source code are scattered around many
files, and you need to group them into one place.<br />
You are responsible for ensuring that the basic template document
(into which individual constant documentation is inserted) contains
all the other information you want, but as a convenience autogsdoc
will generate a simple template (which you may then edit) for you
if the file does not exist.
<br />Insertion takes place immediately before the <em>back</em>
element (or if that does not exist, immediately before the end
of the <em>body</em> element) in the template.
</item>
<item><strong>Declared</strong>
Specify where headers are to be documented as being found.<br />
The actual name produced in the documentation is formed by appending
the last component of the header file name to the value of this
default.<br />
If this default is not specified, the full name of the header file
(as supplied on the command line), with the HeaderDirectory
default prepended, is used.<br />
A typical usage of this might be <code>-Declared Foundation</code>
when generating documentation for the GNUstep base library. This
would result in the documentation saying that NSString is declared
in <code>Foundation/NSString.h</code>
</item>
<item><strong>DocumentAllInstanceVariables</strong>
This flag permits you to generate documentation for all instance
variables. Normally, only those explicitly declared 'public' or
'protected' will be documented.
</item>
<item><strong>DocumentationDirectory</strong>
May be used to specify the directory in which generated
documentation is to be placed. If this is not set, output
is placed in the current directory. This directory is also
used as a last resort to locate source files (not headers).
</item>
<item><strong>Files</strong>
Specifies the name of a file containing a list of file names as
a property list array <em>(name1,name2,...)</em> format. If this
is present, filenames in the program argument list are ignored and
the names in this file are used as the list of names to process.
</item>
<item><strong>FunctionsTemplate</strong>
Specify the name of a template document into which documentation
about functions should be inserted from all files in the project.<br />
This is useful if function source code is scattered around many
files, and you need to group it into one place.<br />
You are responsible for ensuring that the basic template document
(into which individual function documentation is inserted) contains
all the other information you want, but as a convenience autogsdoc
will generate a simple template (which you may then edit) for you
if the file does not exist.
<br />Insertion takes place immediately before the <em>back</em>
element (or if that does not exist, immediately before the end
of the <em>body</em> element) in the template.
</item>
<item><strong>GenerateHtml</strong>
May be used to specify if HTML output is to be generated.
Defaults to YES.
</item>
<item><strong>HeaderDirectory</strong>
May be used to specify the directory to be searched for header files.
When supplied, this value is prepended to relative header names,
otherwise the relative header names are interpreted relative to
the current directory.<br />
Header files specified as absolute paths are not influenced by this
default.
</item>
<item><strong>IgnoreDependencies</strong>
A boolean value which may be used to specify that the program should
ignore file modification times and regenerate files anyway. Provided
for use in conjunction with the <code>make</code> system, which is
expected to manage dependency checking itsself.
</item>
<item><strong>LocalProjects</strong>
This value is used to control the automatic inclusion of local
external projects into the indexing system for generation of
cross-references in final document output.<br />
If set to 'None', then no local project references are done,
otherwise, the 'Local' GNUstep documentation directory is recursively
searched for files with a <code>.igsdoc</code> extension, and the
indexing information from those files is used.<br />
The value of this string is also used to generate the filenames in
the cross reference ... if it is an empty string, the path to use
is assumed to be a file in the same directory where the igsdoc
file was found, otherwise it is used as a prefix to the name in
the index.<br />
NB. Local projects with the same name as the project currently
being documented will <em>not</em> be included by this mechanism.
If you wish to include such projects, you must do so explicitly
using <em>-Projects</em>
</item>
<item><strong>MacrosTemplate</strong>
Specify the name of a template document into which documentation
about macros should be inserted from all files in the project.<br />
This is useful if macro code is scattered around many
files, and you need to group it into one place.<br />
You are responsible for ensuring that the basic template document
(into which individual macro documentation is inserted) contains
all the other information you want, but as a convenience autogsdoc
will generate a simple template (which you may then edit) for you
if the file does not exist.
<br />Insertion takes place immediately before the <em>back</em>
element (or if that does not exist, immediately before the end
of the <em>body</em> element) in the template.
</item>
<item><strong>Project</strong>
May be used to specify the name of this project ... determines the
name of the index reference file produced as part of the documentation
to provide information enabling other projects to cross-reference to
items in this project.
</item>
<item><strong>Projects</strong>
This value may be supplied as a dictionary containing the paths to
the igsdoc index/reference files used by external projects, along
with values to be used to map the filenames found in the indexes.<br />
For example, if a project index (igsdoc) file says that the class
<code>Foo</code> is found in the file <code>Foo</code>, and the
path associated with that project index is <code>/usr/doc/proj</code>,
Then generated html output may reference the class as being in
<code>/usr/doc/prj/Foo.html</code>
</item>
<item><strong>ShowDependencies</strong>
A boolean value which may be used to specify that the program should
log which files are being regenerated because of their dependencies
on other files.
</item>
<item><strong>Standards</strong>
A boolean value used to specify whether the program should insert
information about standards complience into ythe documentation.
This should only be used when documenting the GNUstep libraries
and tools themselves as it assumes that the code being documented
is part of GNUstep and possibly complies with the OpenStep standard
or implements MacOS-X compatible methods.
</item>
<item><strong>SystemProjects</strong>
This value is used to control the automatic inclusion of system
external projects into the indexing system for generation of
cross-references in final document output.<br />
If set to 'None', then no system project references are done,
otherwise, the 'System' GNUstep documentation directory is recursively
searched for files with a <code>.igsdoc</code> extension, and the
indexing information from those files is used.<br />
The value of this string is also used to generate the filenames in
the cross reference ... if it is an empty string, the path to use
is assumed to be a file in the same directory where the igsdoc
file was found, otherwise it is used as a prefix to the name in
the index.<br />
NB. System projects with the same name as the project currently
being documented will <em>not</em> be included by this mechanism.
If you wish to include such projects, you must do so explicitly
using <em>-Projects</em>
</item>
<item><strong>TypedefsTemplate</strong>
Specify the name of a template document into which documentation
about typedefs should be inserted from all files in the project.<br />
This is useful if typedef source code is scattered around many
files, and you need to group it into one place.<br />
You are responsible for ensuring that the basic template document
(into which individual typedef documentation is inserted) contains
all the other information you want, but as a convenience autogsdoc
will generate a simple template (which you may then edit) for you
if the file does not exist.
<br />Insertion takes place immediately before the <em>back</em>
element (or if that does not exist, immediately before the end
of the <em>body</em> element) in the template.
</item>
<item><strong>Up</strong>
A string used to supply the name to be used in the 'up' link from
generated gsdoc documents. This should normally be the name of a
file which contains an index of the contents of a project.<br />
If this is missing or set to an empty string, then no 'up' link
will be provided in the documents.
</item>
<item><strong>VariablesTemplate</strong>
Specify the name of a template document into which documentation
about variables should be inserted from all files in the project.<br />
This is useful if variable source code is scattered around many
files, and you need to group it into one place.<br />
You are responsible for ensuring that the basic template document
(into which individual variable documentation is inserted) contains
all the other information you want, but as a convenience autogsdoc
will generate a simple template (which you may then edit) for you
if the file does not exist.
<br />Insertion takes place immediately before the <em>back</em>
element (or if that does not exist, immediately before the end
of the <em>body</em> element) in the template.
</item>
<item><strong>Verbose</strong>
A boolean used to specify whether you want verbose debug/warning
output to be produced.
</item>
<item><strong>Warn</strong>
A boolean used to specify whether you want standard warning
output (eg report of undocumented methods) produced.
</item>
<item><strong>WordMap</strong>
This value is a dictionary used to map identifiers/keywords found
in the source files to other words. Generally you will not have
to use this, but it is sometimes helpful to avoid the parser being
confused by the use of C preprocessor macros. You can effectively
redefine the macro to something less confusing.<br />
The value you map the identifier to must be one of -<br />
Another identifier,<br />
An empty string - the value is ignored,<br />
Two slashes ('//') - the rest of the line is ignored.<br />
</item>
</list>
<section>
<heading>overview</heading>
<p>
The autogsdoc tool is a command-line utility for parsing ObjectiveC
source code (header files and optionally source files) in order to
generate documentation covering the public interface of the various
classes, categories, and protocols in the source.
</p>
<p>
The simple way to use this is to run the command with one or more
header file names as arguments ... the tool will automatically
parse corresponding source files in the same directory as the
headers (or the current directory, or the directory specified
using the DocumentationDirectory default), and produce gsdoc
and html files as output.
</p>
<p>
Even without any human assistance, this tool will produce skeleton
documents listing the methods in the classes found in the source
files, but more importantly it can take specially formatted comments
from the source files and insert those comments into the gsdoc output.
</p>
<p>
Any comment beginning with slash and <em>two</em> asterisks rather than
the common slash and single asterisk, is taken to be gsdoc markup, to
be use as the description of the class or method following it. This
comment text is reformatted and then inserted into the output.<br />
Where multiple comments are associatd with the same item, they are
joined together with a line break (&lt;br /&gt;) between each if
necessary.
</p>
<p>
The tool can easily be used to document programs as well as libraries,
simply by giving it the name of the source file containing the main()
function of the program - it takes the special comments from that
function and handles them specially, inserting them as a section at
the end of the first chapter of the document (it creates the first
chapter if necessary).
</p>
</section>
<section>
<heading>Extra markup</heading>
<p>
There are some cases where special extra processing is performed,
predominantly in the first comment found in the source file,
from which various chunks of gsdoc markup may be extracted and
placed into appropriate locations in the output document -
</p>
<list>
<item><strong>AutogsdocSource</strong>
In any line where <code>AutogsdocSource</code>: is found, the remainder
of the line is taken as a source file name to be used instead of
making the assumption that each .h file processed uses a .m file
of the same name. You may supply multiple <code>AutogsdocSource</code>:
lines where a header file declares items which are defined in
multiple source files.<br />
If a file name is absolute, it is used just as supplied.<br />
If on the other hand, it is a relative path, the software looks for
the source file first relative to the location of the header file,
and if not found there, relative to the current directory in which
autogsdoc is running, and finally relative to the directory
specified by the DocumentationDirectory default.
</item>
<item><strong>&lt;abstract&gt;</strong>
An abstract of the content of the document ... placed in the head
of the gsdoc output.
</item>
<item><strong>&lt;author&gt;</strong>
A description of the author of the code - may be repeated to handle
the case where a document has multiple authors. Placed in the
head of the gsdoc output.<br />
As an aid to readability of the source, some special additional
processing is performed related to the document author -<br />
Any line of the form '<code>Author</code>: name &lt;email-address&gt;',
or '<code>By</code>: name &lt;email-address&gt;',
or '<code>Author</code>: name' or '<code>By</code>: name'
will be recognised and converted to an <em>author</em> element,
possibly containing an <em>email</em> element.
</item>
<item><strong>&lt;back&gt;</strong>
Placed in the gsdoc output just before the end of the body of the
document - intended to be used for appendices, index etc.
</item>
<item><strong>&lt;chapter&gt;</strong>
Placed immediately before any generated class documentation ...
intended to be used to provide overall description of how the
code being documented works.<br />Any documentation for the main()
function of a program is inserted as a section at the end of this
chapter.
</item>
<item><strong>&lt;copy&gt;</strong>
Copyright of the content of the document ... placed in the head
of the gsdoc output.<br />
As an aid to readability of the source, some special additional
processing is performed -<br />
Any line of the form 'Copyright (C) text' will be recognised and
converted to a <em>copy</em> element.
</item>
<item><strong>&lt;date&gt;</strong>
Date of the revision of the document ... placed in the head
of the gsdoc output. If this is omitted the tool will try to
construct a value from the RCS Date tag (if available).
</item>
<item><strong>&lt;front&gt;</strong>
Inserted into the document at the start of the body ... intended
to provide for introduction or contents pages etc.
</item>
<item><strong>&lt;title&gt;</strong>
Title of the document ... placed in the head of the gsdoc output.
If this is omitted the tool will generate a (probably poor)
title of its own - so you should include this markup manually.
</item>
<item>
<strong>NB</strong>This markup may be used within
class, category, or protocol documentation ... if so, it is
extracted and wrapped round the rest of the documentation for
the class as the classes chapter.
The rest of the class documentation is normally
inserted at the end of the chapter, but may instead be substituted
in in place of the &lt;unit /&gt; pseudo-element within the
&lt;chapter&gt; element.
</item>
<item><strong>&lt;version&gt;</strong>
Version identifier of the document ... placed in the head
of the gsdoc output. If this is omitted the tool will try to
construct a value from the RCS Revision tag (if available).
</item>
</list>
</section>
<section>
<heading>Method markup</heading>
<p>
In comments being used to provide text for a method description, the
following markup is removed from the text and handled specially -
</p>
<list>
<item><strong>&lt;init /&gt;</strong>
The method is marked as being the designated initialiser for the class.
</item>
<item><strong>&lt;override-subclass /&gt;</strong>
The method is marked as being one which subclasses must override
(eg an abstract method).
</item>
<item><strong>&lt;override-never /&gt;</strong>
The method is marked as being one which subclasses should <em>NOT</em>
override.
</item>
<item><strong>&lt;standards&gt; ... &lt;/standards&gt;</strong>
The markup is removed from the description and placed <em>after</em>
it in the gsdoc output - so that the method is described as
conforming (or not conforming) to the specified standards.
</item>
</list>
</section>
<section>
<heading>Automated markup</heading>
<p>
Generally, the text in comments is reformatted to standardise and
indent it nicely ... the reformatting is <em>not</em> performed on
any text inside an &lt;example&gt; element.<br />
When the text is reformatted, it is broken into whitespace separated
'words' which are then subjected to some extra processing ...
</p>
<list>
<item>Certain well known constants such as YES, NO, and nil are
enclosed in &lt;code&gt; ... &lt;/code&gt; markup.
</item>
<item>The names of method arguments within method descriptions are
enclosed in &lt;var&gt; ... &lt;/var&gt; markup.
</item>
<item>Method names (beginning with a plus or minus) are enclosed
in &lt;ref...&gt; ... &lt;/ref&gt; markup.<br />
eg. "-init" (without the quotes) would be wrapped in a gsdoc
reference element to point to the init method of the current
class or, if only one known class had an init method, it
would refer to the method of that class.
<br />Note the fact that the method name must be surrounded by
whitespace to be recognized (though a comma, fullstop, or semicolon
at the end of the specifier will act like whitespace).
</item>
<item>Method specifiers including class names (beginning and ending with
square brackets) are enclosed in &lt;ref...&gt; ... &lt;/ref&gt; markup.
<br />eg. <code>[</code>NSObject-init<code>]</code>,
will create a reference to the init method of NSObject (either the
class proper, or any of its categories), while
<br /><code>[</code>(NSCopying)-copyWithZone:<code>]</code>, creates a
reference to a method in the NSCopyIng protocol.
<br />Note that no spaces must appear between the square brackets
in these specifiers.
<br />Protocol names are enclosed in round brackets rather than
the customary angle brackets, because gsdoc is an XML language, and
XML treats angle brackets specially.
</item>
<item>Class names (and also protocol and category names) enclosed
in square brackets are also cross referenced.
<br />Protocol names are enclosed in round brackets rather than
the customary angle brackets, because gsdoc is an XML language, and
XML treats angle brackets specially.
</item>
<item>Function names (ending with '()') other than 'main()' are enclosed
in &lt;ref...&gt; ... &lt;/ref&gt; markup.<br />
eg. "NSLogv()" (without the quotes) would be wrapped in a gsdoc
reference element to point to the documentation of the NSLog function.
<br />Note the fact that the function name must be surrounded by
whitespace (though a comma, fullstop, or semicolon at the end
of the specifier will also act as a whitespace terminator).
</item>
</list>
</section>
<section>
<heading>Arguments and Defaults</heading>
<p>
The tools accepts certain user defaults (which can of course be
supplied as command-line arguments as usual) -
</p>
<list>
<item><strong>Clean</strong>
If this boolean value is set to YES, then rather than generating
documentation, the tool removes all gsdoc files generated in the
project, and all html files generated from them (as well as any
which would be generated from gsdoc files listed explicitly),
and finally removes the project index file.<br />
The only exception to this is that template gsdoc files (ie those
specifield using "-ConstantsTemplate ...", "-FunctionsTemplate ..."
arguments etc) are not deleted unless the CleanTemplates flag is set.
</item>
<item><strong>CleanTemplates</strong>
This flag specifies whether template gsdoc files are to be removed
along with other files when the Clean option is specified.
The default is for them not to be removed ... since these templates
may have been produced manually and just had data inserted into them.
</item>
<item><strong>ConstantsTemplate</strong>
Specify the name of a template document into which documentation
about constants should be inserted from all files in the project.<br />
This is useful if constants in the source code are scattered around many
files, and you need to group them into one place.<br />
You are responsible for ensuring that the basic template document
(into which individual constant documentation is inserted) contains
all the other information you want, but as a convenience autogsdoc
will generate a simple template (which you may then edit) for you
if the file does not exist.
<br />Insertion takes place immediately before the <em>back</em>
element (or if that does not exist, immediately before the end
of the <em>body</em> element) in the template.
</item>
<item><strong>Declared</strong>
Specify where headers are to be documented as being found.<br />
The actual name produced in the documentation is formed by appending
the last component of the header file name to the value of this
default.<br />
If this default is not specified, the full name of the header file
(as supplied on the command line), with the HeaderDirectory
default prepended, is used.<br />
A typical usage of this might be <code>"-Declared Foundation"</code>
when generating documentation for the GNUstep base library. This
would result in the documentation saying that NSString is declared
in <code>Foundation/NSString.h</code>
</item>
<item><strong>DocumentAllInstanceVariables</strong>
This flag permits you to generate documentation for all instance
variables. Normally, only those explicitly declared 'public' or
'protected' will be documented.
</item>
<item><strong>DocumentationDirectory</strong>
May be used to specify the directory in which generated
documentation is to be placed. If this is not set, output
is placed in the current directory. This directory is also
used as a last resort to locate source files (not headers).
</item>
<item><strong>Files</strong>
Specifies the name of a file containing a list of file names as
a property list array <em>(name1,name2,...)</em> format. If this
is present, filenames in the program argument list are ignored and
the names in this file are used as the list of names to process.
</item>
<item><strong>FunctionsTemplate</strong>
Specify the name of a template document into which documentation
about functions should be inserted from all files in the project.<br />
This is useful if function source code is scattered around many
files, and you need to group it into one place.<br />
You are responsible for ensuring that the basic template document
(into which individual function documentation is inserted) contains
all the other information you want, but as a convenience autogsdoc
will generate a simple template (which you may then edit) for you
if the file does not exist.
<br />Insertion takes place immediately before the <em>back</em>
element (or if that does not exist, immediately before the end
of the <em>body</em> element) in the template.
</item>
<item><strong>GenerateHtml</strong>
May be used to specify if HTML output is to be generated.
Defaults to YES.
</item>
<item><strong>HeaderDirectory</strong>
May be used to specify the directory to be searched for header files.
When supplied, this value is prepended to relative header names,
otherwise the relative header names are interpreted relative to
the current directory.<br />
Header files specified as absolute paths are not influenced by this
default.
</item>
<item><strong>IgnoreDependencies</strong>
A boolean value which may be used to specify that the program should
ignore file modification times and regenerate files anyway. Provided
for use in conjunction with the <code>make</code> system, which is
expected to manage dependency checking itsself.
</item>
<item><strong>LocalProjects</strong>
This value is used to control the automatic inclusion of local
external projects into the indexing system for generation of
cross-references in final document output.<br />
If set to 'None', then no local project references are done,
otherwise, the 'Local' GNUstep documentation directory is recursively
searched for files with a <code>.igsdoc</code> extension, and the
indexing information from those files is used.<br />
The value of this string is also used to generate the filenames in
the cross reference ... if it is an empty string, the path to use
is assumed to be a file in the same directory where the igsdoc
file was found, otherwise it is used as a prefix to the name in
the index.<br />
NB. Local projects with the same name as the project currently
being documented will <em>not</em> be included by this mechanism.
If you wish to include such projects, you must do so explicitly
using <em>"-Projects ..."</em>
</item>
<item><strong>MacrosTemplate</strong>
Specify the name of a template document into which documentation
about macros should be inserted from all files in the project.<br />
This is useful if macro code is scattered around many
files, and you need to group it into one place.<br />
You are responsible for ensuring that the basic template document
(into which individual macro documentation is inserted) contains
all the other information you want, but as a convenience autogsdoc
will generate a simple template (which you may then edit) for you
if the file does not exist.
<br />Insertion takes place immediately before the <em>back</em>
element (or if that does not exist, immediately before the end
of the <em>body</em> element) in the template.
</item>
<item><strong>Project</strong>
May be used to specify the name of this project ... determines the
name of the index reference file produced as part of the documentation
to provide information enabling other projects to cross-reference to
items in this project.
</item>
<item><strong>Projects</strong>
This value may be supplied as a dictionary containing the paths to
the igsdoc index/reference files used by external projects, along
with values to be used to map the filenames found in the indexes.<br />
For example, if a project index (igsdoc) file says that the class
<code>Foo</code> is found in the file <code>Foo</code>, and the
path associated with that project index is <code>/usr/doc/proj</code>,
Then generated html output may reference the class as being in
<code>/usr/doc/prj/Foo.html</code>
</item>
<item><strong>ShowDependencies</strong>
A boolean value which may be used to specify that the program should
log which files are being regenerated because of their dependencies
on other files.
</item>
<item><strong>Standards</strong>
A boolean value used to specify whether the program should insert
information about standards complience into ythe documentation.
This should only be used when documenting the GNUstep libraries
and tools themselves as it assumes that the code being documented
is part of GNUstep and possibly complies with the OpenStep standard
or implements MacOS-X compatible methods.
</item>
<item><strong>SystemProjects</strong>
This value is used to control the automatic inclusion of system
external projects into the indexing system for generation of
cross-references in final document output.<br />
If set to 'None', then no system project references are done,
otherwise, the 'System' GNUstep documentation directory is recursively
searched for files with a <code>.igsdoc</code> extension, and the
indexing information from those files is used.<br />
The value of this string is also used to generate the filenames in
the cross reference ... if it is an empty string, the path to use
is assumed to be a file in the same directory where the igsdoc
file was found, otherwise it is used as a prefix to the name in
the index.<br />
NB. System projects with the same name as the project currently
being documented will <em>not</em> be included by this mechanism.
If you wish to include such projects, you must do so explicitly
using <em>"-Projects ..."</em>
</item>
<item><strong>TypedefsTemplate</strong>
Specify the name of a template document into which documentation
about typedefs should be inserted from all files in the project.<br />
This is useful if typedef source code is scattered around many
files, and you need to group it into one place.<br />
You are responsible for ensuring that the basic template document
(into which individual typedef documentation is inserted) contains
all the other information you want, but as a convenience autogsdoc
will generate a simple template (which you may then edit) for you
if the file does not exist.
<br />Insertion takes place immediately before the <em>back</em>
element (or if that does not exist, immediately before the end
of the <em>body</em> element) in the template.
</item>
<item><strong>Up</strong>
A string used to supply the name to be used in the 'up' link from
generated gsdoc documents. This should normally be the name of a
file which contains an index of the contents of a project.<br />
If this is missing or set to an empty string, then no 'up' link
will be provided in the documents.
</item>
<item><strong>VariablesTemplate</strong>
Specify the name of a template document into which documentation
about variables should be inserted from all files in the project.<br />
This is useful if variable source code is scattered around many
files, and you need to group it into one place.<br />
You are responsible for ensuring that the basic template document
(into which individual variable documentation is inserted) contains
all the other information you want, but as a convenience autogsdoc
will generate a simple template (which you may then edit) for you
if the file does not exist.
<br />Insertion takes place immediately before the <em>back</em>
element (or if that does not exist, immediately before the end
of the <em>body</em> element) in the template.
</item>
<item><strong>Verbose</strong>
A boolean used to specify whether you want verbose debug/warning
output to be produced.
</item>
<item><strong>Warn</strong>
A boolean used to specify whether you want standard warning
output (eg report of undocumented methods) produced.
</item>
<item><strong>WordMap</strong>
This value is a dictionary used to map identifiers/keywords found
in the source files to other words. Generally you will not have
to use this, but it is sometimes helpful to avoid the parser being
confused by the use of C preprocessor macros. You can effectively
redefine the macro to something less confusing.<br />
The value you map the identifier to must be one of -<br />
Another identifier,<br />
An empty string - the value is ignored,<br />
Two slashes ('//') - the rest of the line is ignored.<br />
</item>
</list>
</section>
<section>
<heading>Inter-document linkage</heading>
<p>

View file

@ -149,7 +149,7 @@
<!ELEMENT ref (%text;)*>
<!ATTLIST ref
id CDATA #REQUIRED
type (class|protocol|method|ivariable|function|type|macro|variable|constant|label|EOModel|EOEntity) "label"
type (class|category|protocol|method|ivariable|function|type|macro|variable|constant|label|EOModel|EOEntity) "label"
class CDATA #IMPLIED
>