gs/doc/C-style.htm

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd">
<html>
<head>
<title>Ghostscript C coding guidelines</title>
<!-- $Id: C-style.htm,v 1.21.2.2 2002/02/01 05:31:24 raph Exp $ -->
<!-- Originally: c-style.txt -->
<link rel="stylesheet" type="text/css" href="gs.css" title="Ghostscript Style">
</head>

<body>
<!-- [1.0 begin visible header] ============================================ -->

<!-- [1.1 begin headline] ================================================== -->

<h1>Ghostscript C coding guidelines</h1>

<!-- [1.1 end headline] ==================================================== -->

<!-- [1.2 begin table of contents] ========================================= -->

<h2>Table of contents</h2>

<blockquote><ul>
<li><a href="#Introduction">Introduction</a>
<li><a href="#C_language">C language do's and don'ts</a>
<ul>
<li>Preprocessor:
    <a href="#Conditionals">Conditionals</a>,
    <a href="#Macros">Macros</a>,
    <a href="#Preprocessor_other">Other</a>
<li><a href="#Lexical_elements">Lexical elements</a>
<li><a href="#Scoping">Scoping</a>
<li>Data types:
    <a href="#Scalars">Scalars</a>,
    <a href="#Arrays">Arrays</a>,
    <a href="#Typedefs">Typedefs</a>,
    <a href="#Structures">Structures</a>,
    <a href="#Unions">Unions</a>
<li><a href="#Expressions">Expressions</a>
<li><a href="#Statements">Statements</a>
<li><a href="#Procedures">Procedures</a> (prototypes and definitions)
<li><a href="#Standard_library">Standard library</a>
</ul>
<li><a href="#Language_extensions">Language extensions</a>
<li><a href="#Stylistic_conventions">Stylistic conventions</a>
<ul>
<li>Formatting:
    <a href="#Indentation">Indentation</a>,
    <a href="#Spaces">Spaces</a>,
    <a href="#Parentheses">Parentheses</a>
<li><a href="#Preprocessor_style">Preprocessor</a>
<li><a href="#Naming">Naming</a>
<li><a href="#Types">Types</a>
<li><a href="#Procedures_style">Procedures</a>,
<li>Miscellany:
    <a href="#Local_variables">Local variables</a>,
    <a href="#Compiler_warnings">Compiler warnings</a>
</ul>
<li><a href="#File_structuring">File structuring and naming</a>
<ul>
<li><a href="#All_files">All files</a>
<li><a href="#Makefiles">Makefiles</a>
<li><a href="#General_C_code">General C Code</a>
<li><a href="#Headers">Headers (<b><tt>.h</tt></b> files)</a>
<li><a href="#Source">Source (<b><tt>.c</tt></b> files)</a>
</ul>
<li><a href="#Conventions">Ghostscript conventions</a>
<ul>
<li><a href="#Specific_names">Specific names</a>:
    <a href="#code"><b><tt>code</tt></b></a>,
    <a href="#status"><b><tt>status</tt></b></a>
<li><a href="#Structure_type_descriptors">Structure type descriptors</a>
<li><a href="#Objects">"Objects"</a>
<li><a href="#Error_handling">Error handling</a>
</ul>
</ul></blockquote>

<!-- [1.2 end table of contents] =========================================== -->

<!-- [1.3 begin hint] ====================================================== -->

<p>
For other information, see the <a href="Readme.htm">Ghostscript
overview</a>.

<!-- [1.3 end hint] ======================================================== -->

<hr>

<!-- [1.0 end visible header] ============================================== -->

<!-- [2.0 begin contents] ================================================== -->

<h2><a name="Introduction"></a>Introduction</h2>

<p>
This document describes Ghostscript's C coding conventions.  It is primarily
<em>prescriptive</em>, documenting what developers should do when writing
new code; the companion developer documentation (<a
 href="Develop.htm">Develop.htm</a>) is primarily <em>descriptive</em>,
documenting the way things are.

<p>
We encourage following the general language usage and stylistic rules for
any code that will be integrated with Ghostscript, even if the code doesn't
use Ghostscript's run-time facilities or have anything to do with
PostScript, PDF, or page description languages.  Ghostscript itself follows
some additional conventions; these are documented separately under "<a
href="#Conventions">Ghostscript conventions</a>" below.

<hr>

<h2><a name="C_language"></a>C language do's and don'ts</h2>

<p>
There are several different versions of the C language, and even of the ANSI
C standard.  Ghostscript versions through 7.0 were designed to compile on
pre-ANSI compilers as well as on many compilers that implemented the ANSI
standard with varying faithfulness.  Ghostscript versions since 7.0 do not
cater for pre-ANSI compilers: they must conform to the ANSI 1989 standard
(ANS X3.159-1989), with certain restrictions and a few conventional
additions.

<h3>Preprocessor</h3>

<h4><a name="Conditionals"></a>Conditionals</h4>

Restrictions:

<ul>

<li>Don't assume that <b><tt>#if</tt></b> will treat undefined names as 0.
While the ANSI standard requires this, it may produce a warning.

<li>In <b><tt>.c</tt></b> files, don't use preprocessor conditionals that
test for individual platforms or compilers.  Use them only in header files
named xxx<b><tt>_.h</tt></b>.

</ul>

<h4><a name="Macros"></a>Macros</h4>

<p>
Restrictions:

<ul>

<li>Don't redefine a macro, even with the same definition, without using
<b><tt>#undef</tt></b>.

<li><b><tt>CAPITALIZE</tt></b> macro names unless there is a good reason not
to.

<li>Don't use a macro call within a macro argument if the call expands to a
token sequence that includes any commas not within parentheses: this
produces different results depending on whether the compiler expands the
inner call before or after the argument is substituted into the macro body.
(The ANSI standard says that calls must be expanded after substitution, but
some compilers do it the other way.)

<li>Don't use macro names, even inadvertently, in string constants.  Some
compilers erroneously try to expand them.

<li>Don't use macros to define shorthands for casted pointers.  For
instance, avoid

<blockquote><b><tt>
#define fdev ((gx_device_fubar *)dev)
</tt></b></blockquote>

<p>
and instead use

<blockquote><b><tt>
gx_device_fubar * const fdev = (gx_device_fubar *)dev;
</tt></b></blockquote>

<p>
The use of <b><tt>const</tt></b> alerts the reader that this is effectively
a synonym.

<li>If a macro generates anything larger than a single expression (that is,
one or more statements), surround it with <b><tt>BEGIN</tt></b> and
<b><tt>END</tt></b>.  These work around the fact that simple statements and
compound statements in C can't be substituted for each other syntactically.

</ul>

<h3><a name="Preprocessor_other"></a>Other</h3>

<p>
Restrictions:

<ul>

<li>Only use <b><tt>#pragma</tt></b> in files that are explicitly identified
as being platform-dependent.  Many compilers complain if this is used at
all, and some complain if they don't recognize the specific pragma being
requested (both incorrect according to the ANSI standard).

</ul>

<h3><a name="Lexical_elements"></a>Lexical elements</h3>

<p>
Do not use:

<ul>

<li>ANSI trigraphs (??x)
<li>Nested comments (/* /* */ */) (not ANSI compliant, but often accepted)
<li>Multi-character character constants ('abc')
<li>Wide-character character or string constants (L'x', L"x")

</ul>

<p>
Restrictions:

<ul>

<li>Procedure and static variable names must be 31 characters or less.

<li>Externally visible procedure and variable names must be unique in the
first 23 characters.

</ul>

<h3><a name="Scoping"></a>Scoping (extern, static, ...)</h3>

<p>
Do not use:

<ul>

<li><b><tt>register</tt></b>

</ul>

<p>
Restrictions:

<ul>

<li>Do not allow a global variable (constant) to have more than one
non-<b><tt>extern</tt></b> definition, even though some ANSI C compilers
allow this.  Every global constant should have exactly one definition, in a
<b><tt>.c</tt></b> file, and preferably just one <b><tt>extern</tt></b>
declaration, in a header file.

<li>Use <b><tt>private</tt></b> instead of <b><tt>static</tt></b> for
procedures and variables declared at the outermost scope of a file.  This
allows making such constructs either visible or invisible to profilers by
changing a single <b><tt>#define</tt></b>.

<li><b><tt>static</tt></b> or <b><tt>private</tt></b> variables must be
<b><tt>const</tt></b> and initialized: non-<b><tt>const</tt></b> statically
allocated variables are incompatible with reentrancy, and we're in the
process of eliminating all of them.

<li>Do not use <b><tt>extern</tt></b> in <b><tt>.c</tt></b> files, only in
<b><tt>.h</tt></b> files, unless you have a very good reason for it (e.g.,
as in <a href="../src/iconf.c">iconf.c</a>).  There are too many such
<b><tt>extern</tt></b>s in the code now: we are eliminating them over time.

<li>Do not declare the same name as both <b><tt>static</tt></b>
(<b><tt>private</tt></b>) and non-<b><tt>static</tt></b> within the same
compilation.  (Some compilers complain, some do not.)  This is especially a
problem for procedures: it is easy to declare a procedure as
<b><tt>private</tt></b> near the beginning of a file and accidentally not
declare it <b><tt>private</tt></b> where it is defined later in the file.

<li>Even though the ANSI standard allows initialized external declarations
(<b><tt>extern&nbsp;int&nbsp;x&nbsp;=&nbsp;0</tt></b>), don't use them.

</ul>

<h3><a name="Scalars"></a>Scalars</h3>

<p>
Restrictions:

<ul>

<li>Avoid using <b><tt>char</tt></b>, except for <b><tt>char&nbsp;*</tt></b>
for a pointer to a string.  Don't assume that <b><tt>char</tt></b> is
signed; also don't assume it is unsigned.

<li>Never cast a <b><tt>float</tt></b> to a <b><tt>double</tt></b>
explicitly.  ANSI compilers in their default mode do all floating point
computations in double precision, and handle such casts automatically.

<li>Don't use <b><tt>long long</tt></b>: even though it is in the ANSI
standard, not all compilers support it.  Use <b><tt>bits64</tt></b> instead
(see below under "<a href="#Language_extensions">Language extensions</a>").

<li>Don't assume anything about whether <b><tt>sizeof(long)</tt></b> is less
than, equal to, or greater than <b><tt>sizeof(ptr)</tt></b>.  (However, you
can make such comparisons in preprocessor conditionals using
<b><tt>ARCH_SIZEOF_LONG</tt></b> and <b><tt>ARCH_SIZEOF_PTR</tt></b>.)

</ul>

<h3><a name="Arrays"></a>Arrays</h3>

<p>
Restrictions:

<ul>

<li>Don't declare arrays of size 0.  (The newer ANSI standard allows this,
but the older one doesn't.)

<li>Don't declare an array of size 1 at the end of a structure to indicate
that a variable-size array follows.

<li>Don't declare initialized <b><tt>auto</tt></b> arrays.

</ul>

<h3><a name="Typedefs"></a>Typedefs</h3>

<p>
Restrictions:

<ul>

<li>Don't use <b><tt>typedef</tt></b> for function types, such as

<blockquote>
<b><tt>typedef int proc_xyz_t(double, int *);</tt></b>
</blockquote>

<p>Many compilers don't handle this correctly -- they will give errors, or
do the wrong thing, when declaring variables of type
<b><tt>proc_xyz_t</tt></b> and/or <b><tt>proc_xyz_t *</tt></b>.  Instead, do
this:

<blockquote>
<b><tt>#define PROC_XYZ(proc) int proc(double, int *)<br>
PROC_XYZ(some_proc);  /* declare a procedure of this type */<br>
typedef PROC_XYZ((*proc_xyz_ptr_t));  /* define a type for procedure ptrs */<br>
<br>
proc_xyz_ptr_t pp;  /* pointer to procedure */</tt></b>
</blockquote>

<li>Don't redefine <b><tt>typedef</tt></b>'ed names, even with the same
definition.  Some compilers complain about this, and the standard doesn't
allow it.

</ul>

<h3><a name="Structures"></a>Structures</h3>

<p>
Restrictions:

<li>Don't use anonymous structures if you can possibly avoid it, except
occasionally as components of other structures. Ideally, use the
<b><tt>struct</tt></b> keyword only for declaring named structure types,
like this:

<blockquote>
<b><tt>typedef struct xxx_s {</tt></b><br>
&nbsp;&nbsp;&nbsp;... members ...<br>
<b><tt>} xxx_t;</tt></b>
</blockquote>

<li>Use <b><tt>struct</tt></b> only when declaring structure types, never
for referring to them (e.g., never declare a variable as type
<b><tt>struct&nbsp;xxx_s&nbsp;*</tt></b>).

<li>Don't assume that the compiler will (or won't) insert padding in
structures to align members for best performance.  To preserve alignment,
only declare structure members that are narrower than an <b><tt>int</tt></b>
if there will be a lot of instances of that structure in memory.  For such
structures, insert <b><tt>byte</tt></b> and/or <b><tt>short</tt></b> padding
members as necessary to re-establish <b><tt>int</tt></b> alignment.

<li>Don't declare initialized <b><tt>auto</tt></b> structures.

</ul>

<h3><a name="Unions"></a>Unions</h3>

<p>
Restrictions:

<ul>

<li>Use unions only as components of structures, not as typedefs in their
own right.

<li>Don't attempt to initialize unions: not all compilers support this, even
though it is in the 1989 ANSI standard.

</ul>

<h3><a name="Expressions"></a>Expressions</h3>

<p>
Restrictions:

<ul>

<li>Don't assign a larger integer data type to a smaller one without a cast
(<b><tt>int_x&nbsp;=&nbsp;long_y</tt></b>).

<li>It's OK to use the address of a structure or array element
(<b><tt>&p->e</tt></b>, <b><tt>&a[i]</tt></b>) locally, or pass it to a
procedure that won't store it, but don't store such an address in allocated
storage unless you're very sure of what you're doing.

<li>Don't use conditional expressions with structure or union values.
(Pointers to structures or unions are OK.)

<li>For calling a variable or parameter procedure, use
<b><tt>ptr-&gt;func(...)</tt></b>.  Some old code uses explicit indirection,
<b><tt>(*ptr-&gt;func)(...)</tt></b>: don't use this in new code.

<li>Don't write expressions that depend on order of evaluation, unless the
order is created explicitly by use of <b><tt>||</tt></b>,
<b><tt>&amp;&amp;</tt></b>, <b><tt>?:</tt></b>, <b><tt>,</tt></b>, or
function nesting (the arguments of a function must be evaluated before the
function is called).  In particular, don't assume that the arguments of a
function will be evaluated left-to-right, or that the left side of an
assignment will be evaluated before the right.

<li>Don't mix integer and enumerated types ad lib: treat enumerated types as
distinct from integer types, and use casts to convert between the two.
(Some compilers generate warnings if you do not do this.)

</ul>

<h3><a name="Statements"></a>Statements</h3>

<p>
Restrictions:

<ul>

<li>If you use an expression as a statement, other than an assignment or a
function call with <b><tt>void</tt></b> return value, enclose it explicitly
in <b><tt>DISCARD()</tt></b>.

<li>The type of the operand of a <b><tt>switch</tt></b> must match the type
of the case labels, whether the labels are <b><tt>int</tt></b>s or the
members of an <b><tt>enum</tt></b> type.  (Use a cast if necessary.)

<li>It is OK for one case of a switch to "fall through" into another (i.e.,
for the statement just before a case label not to be a control transfer),
but a comment <b><tt>/*&nbsp;falls&nbsp;through&nbsp;*/</tt></b> is
required.

<li>If you are returning an error code specified explicitly (e.g.,
<b><tt>return&nbsp;gs_error_rangecheck</tt></b> or
<b><tt>return&nbsp;e_rangecheck</tt></b>), use
<b><tt>return_error()</tt></b> rather than plain <b><tt>return</tt></b>.
However, if the program is simply propagating an error code generated
elsewhere, as opposed to generating the error, use <b><tt>return</tt></b>
(e.g., <b><tt>if&nbsp;(code&nbsp;<&nbsp;0)&nbsp;return&nbsp;code</tt></b>).

</ul>

<h3><a name="Procedures"></a>Procedures</h3>

<p>
Restrictions:

<ul>

<li>Provide a prototype for every procedure, and make sure the prototype is
available at every call site.  If the procedure is local to a file
(<b><tt>private</tt></b>), the prototype should precede the procedure, in
the same file; if the procedure is global, the prototype should be in a
header file.

<li>If a procedure parameter is itself a procedure, do list its parameter
types rather than just using <b><tt>()</tt></b>.  For example,

<blockquote><b><tt>
int foo(int (*callback)(int, int));
</tt></b></blockquote>

<p>
rather than just

<blockquote><b><tt>
int foo(int (*callback)());
</tt></b></blockquote>

<li>Don't use the <b><tt>P</tt></b>* macros in new code.  (See the
Procedures section of <a href="#Language_extensions">Language extensions</a>
below for more information.)

<li>Always provide an explicit return type for procedures, in both the
prototype and the definition: don't rely on the implicit declaration as
<b><tt>int</tt></b>.

<li>Don't use <b><tt>float</tt></b> as the return type of a procedure,
unless there's a special reason.  Floating point hardware typically does
everything in double precision internally and has to do extra work to
convert between double and single precision.

<li>Don't declare parameters as being of type <b><tt>float</tt></b>,
<b><tt>short</tt></b>, or <b><tt>char</tt></b>.  If you do this and forget
to include the prototype at a call site, ANSI compilers will generate
incompatible calling sequences.  Use <b><tt>floatp</tt></b> (a synonym for
<b><tt>double</tt></b>, mnemonic for "float parameter") instead of
<b><tt>float</tt></b>, and use <b><tt>int</tt></b> or <b><tt>uint</tt></b>
instead of <b><tt>short</tt></b> or <b><tt>char</tt></b>.

</ul>

<h3><a name="Standard_library"></a>Standard library</h3>

<p>
Restrictions:

<ul>

<li>Only use library features that are documented in the established ANSI
standard (e.g., Harbison & Steele's book).  Do not use procedures that are
"standards" promulgated by Microsoft (e.g., <b><tt>stricmp</tt></b>), or
originate in BSD Unix (e.g., <b><tt>strcasecmp</tt></b>), or were added in
later versions of the standard such as C 9X.

<li>Do not use any features from <b><tt>stdio.h</tt></b> that assume the
existence of <b><tt>stdin</tt></b>, <b><tt>stdout</tt></b>, or
<b><tt>stderr</tt></b>.  See <a href="../src/gsio.h">gsio.h</a> for the full
list.  Instead, use <b><tt>gs_stdin</tt></b> et al.

</ul>

<hr>

<h2><a name="Language_extensions"></a>Language extensions</h2>

<h3>Scoping</h3>

<dl>

<dt><b><tt>inline</tt></b>

<dd><b><tt>inline</tt></b> is available even if the compiler does not
support it.  Be aware, however, that it may have no effect.  In particular,
do not use <b><tt>inline</tt></b> in header files.  Instead, use the
<b><tt>extern_inline</tt></b> facility described just below.

<dt><b><tt>extern_inline</tt></b>

<dd>Compilers that do support <b><tt>inline</tt></b> vary in how they decide
whether to (also) compile a closed-code copy of the procedure.  Because of
this, putting an <b><tt>inline</tt></b> procedure in a header file may
produce multiple closed copies, causing duplicate name errors at link time.
<b><tt>extern_inline</tt></b> provides a safe way to put
<b><tt>inline</tt></b> procedures in header files, regardless of compiler.
Unfortunately, the only way we've found to make this fully portable involves
a fair amount of boilerplate.  For details, please see <a
href="../src/stdpre.h">stdpre.h</a>.

<dt><b><tt>private</tt></b>

<dd><b>Use <tt>private</tt></b> instead of <b><tt>static</tt></b> for all
file-local procedures, and also for file-local variables defined at the
outermost level.  However, use <b><tt>static</tt></b>, not
<b><tt>private</tt></b>, for variables defined within a procedure.

<p>
<b><tt>private</tt></b> is normally #define'd as <b><tt>static</tt></b>.
However, it can also be #define'd as empty, which allows profilers to
measure all procedures, and 'nm' to list all interesting statically
allocated variables, not just public ones.

</dl>

<h3>Scalar types</h3>

<dl>

<dt><b><tt>bool, true, false</tt></b>

<dd><b><tt>bool</tt></b> is intended as a Boolean type, with canonical
values <b><tt>true</tt></b> and <b><tt>false</tt></b>.  In a more reasonable
language, such as Java, <b><tt>bool</tt></b> is an enumerated type requiring
an explicit cast to or from <b><tt>int</tt></b>; however, because C's
conditionals are defined as producing <b><tt>int</tt></b> values, we can't
even define <b><tt>bool</tt></b> as a C <b><tt>enum</tt></b> without
provoking compiler warnings.

<p>
Even though <b><tt>bool</tt></b> is a synonym for <b><tt>int</tt></b>, treat
them as conceptually different types:

<ul>
<li>Initialize or set <b><tt>bool</tt></b> variables to <b><tt>true</tt></b>
or <b><tt>false</tt></b>, not 0 or 1.
<li>Use the Boolean operators <b><tt>!</tt></b>, <b><tt>&&</tt></b>,
and <b><tt>||</tt></b> only with Booleans.  Don't use the idiom
<b><tt>!!x</tt></b> to create a Boolean that is true iff <b><tt>x</tt></b>
!= 0: use <b><tt>x != 0</tt></b>.
<li>Use an explicit <b><tt>(int)</tt></b> cast to convert a Boolean to an
integer.
</ul>

<dt><b><tt>byte, ushort, uint, ulong</tt></b>

<dd>These types are simply shorthands for <b><tt>unsigned char, short, int,
long</tt></b>.

<dt><b><tt>floatp</tt></b>

<dd>This is a synonym for <b><tt>double</tt></b>.  It should be used for,
and only for, procedure parameters that would otherwise be
<b><tt>float</tt></b>.  (As noted above, procedure parameters should not be
declared as <b><tt>float</tt></b>.)

<dt><b><tt>bits8, bits16, bits32</tt></b>

<dd>These are unsigned integer types of the given width.  Use them wherever
the actual width matters: do <em>not</em>, for example, use
<b><tt>short</tt></b> assuming that it is 16 bits wide.

<dt><b><tt>bits64</tt></b>

<dd><strong>****** NOT IMPLEMENTED YET ******</strong>
This is an unsigned 64-bit integer type, but it may not be available on all
platforms.  Any code that uses this type should be surrounded by
<b><tt>#if&nbsp;ARCH_HAS_BITS64</tt></b>.

</dl>

<h3>Procedures</h3>

<dl>

<dt><b><tt>P0(), P1(p1), ..., P16(p1, ..., p16)</tt></b> [deprecated]

<dd>Nearly all existing code uses these macros for the parameter lists of
procedure prototypes, e.g.,
<b><tt>int&nbsp;proc(P2(int&nbsp;a,&nbsp;int&nbsp;b))</tt></b>, to provide
compatibility with pre-ANSI, a.k.a. "traditional" or "K &amp; R", compilers.
Since only ANSI compilers are now supported, don't use these macros in new
code: use the ANSI standard syntax
<b><tt>int&nbsp;proc(int&nbsp;a,&nbsp;int&nbsp;b)</tt></b>.

</dl>

<hr>

<h2><a name="Stylistic_conventions"></a>Stylistic conventions</h2>

<p>
Ghostscript's coding rules cover not only the use of the language, but also
many stylistic issues like the choice of names and the use of whitespace.
The stylistic rules are meant to produce code that is easy to read.  It's
important to observe them as much as possible in order to maintain a
consistent style, but if you find these rules getting in your way or
producing ugly-looking results once in a while, it's OK to break it.

<h3><a name="Formatting"></a>Formatting</h3>

<h4><a name="Indentation"></a>Indentation</h4>

<p>
We've formatted all of our code using the GNU <b><tt>indent</tt></b> program.

<blockquote><b><tt>
indent&nbsp;-bad&nbsp;-nbap&nbsp;-nsob&nbsp;-br&nbsp;-ce&nbsp;-cli4&nbsp;-npcs&nbsp;-ncs&nbsp;\<br>
&nbsp;&nbsp;&nbsp;-i4&nbsp;-di0&nbsp;-psl&nbsp;-lp&nbsp;-lps&nbsp;somefile.c
</tt></b></blockquote>

<p>
does a 98% accurate job of producing our preferred style.  Unfortunately,
there are bugs in all versions of GNU <b><tt>indent</tt></b>, requiring
both pre- and post-processing of the code.  The <b><tt>gsindent</tt></b>
script in the Ghostscript fileset contains the necessary workarounds.

<p>
Put indentation points every 4 spaces, with 8 spaces = 1 tab stop.

<p>
For assignments (including chain assignments), put the entire statement on
one line if it will fit; if not, break it after a <b><tt>=</tt></b> and
indent all the following lines.  I.e., format like this:

<blockquote>
var1&nbsp;<b><tt>=</tt></b>&nbsp;value<b><tt>;</tt></b><br>
var1&nbsp;<b><tt>=</tt></b>&nbsp;var2&nbsp;<b><tt>=</tt></b>&nbsp;value<b><tt>;</tt></b><br>
var1&nbsp;<b><tt>=</tt></b><br>
&nbsp;&nbsp;&nbsp;&nbsp;value<b><tt>;</tt></b><br>
var1&nbsp;<b><tt>=</tt></b><br>
&nbsp;&nbsp;&nbsp;&nbsp;var2&nbsp;<b><tt>=</tt></b>&nbsp;value<b><tt>;</tt></b><br>
var1&nbsp;<b><tt>=</tt></b>&nbsp;var2&nbsp;<b><tt>=</tt></b><br>
&nbsp;&nbsp;&nbsp;&nbsp;value<b><tt>;</tt></b>
</blockquote>

<p>
But not like this:

<blockquote>
var1&nbsp;<b><tt>=</tt></b><br>
var2&nbsp;<b><tt>=</tt></b> value<b><tt>;</tt></b>
</blockquote>

<p>
Indent in-line blocks thus:

<blockquote>
<b><tt>{</tt></b><br>
&nbsp;&nbsp;&nbsp;... declarations ...<br>
&nbsp;&nbsp;&nbsp;{{ blank line if any declarations above }}<br>
&nbsp;&nbsp;&nbsp;... statements ...<br>
<b><tt>}</tt></b>
</blockquote>

<p>
Similarly, indent procedures thus:

<blockquote>
return_type<br>
proc_name(... arguments ...)<br>
<b><tt>{</tt></b><br>
&nbsp;&nbsp;&nbsp;... declarations ...<br>
&nbsp;&nbsp;&nbsp;{{ blank line if any declarations above }}<br>
&nbsp;&nbsp;&nbsp;... statements ...<br>
<b><tt>}</tt></b>
</blockquote>

<p>
If a control construct (<b><tt>if</tt></b>, <b><tt>do</tt></b>,
<b><tt>while</tt></b>, or <b><tt>for</tt></b>) has a one-line body, use
this:

<blockquote>
... control construct ...<br>
&nbsp;&nbsp;&nbsp;... subordinate simple statement ...
</blockquote>

<p>
If it has a multi-line body, use this:

<blockquote>
... control construct ... <b><tt>{</tt></b><br>
&nbsp;&nbsp;&nbsp;... subordinate code ...<br>
<b><tt>}</tt></b>
</blockquote>

<p>
If the subordinate code has declarations, see blocks above.

<p>
For if-else statements, do this:

<blockquote>
<b><tt>if (</tt></b> ...<b><tt> ) {</tt></b><br>
&nbsp;&nbsp;&nbsp;... subordinate code ...<br>
<b><tt>} else if (</tt></b> ...<b><tt> ) {</tt></b><br>
&nbsp;&nbsp;&nbsp;... subordinate code ...<br>
<b><tt>} else {</tt></b><br>
&nbsp;&nbsp;&nbsp;... subordinate code ...<br>
<b><tt>}</tt></b>
</blockquote>

<p>
When there are more than two alternatives, as in the example above, use the
above ("parallel") syntax rather than the following ("nested") syntax:

<blockquote>
<b><tt>if (</tt></b> ...<b><tt> ) {</tt></b><br>
&nbsp;&nbsp;&nbsp;... subordinate code ...<br>
<b><tt>} else {</tt></b><br>
<b><tt>&nbsp;&nbsp;&nbsp;if (</tt></b> ...<b><tt> ) {</tt></b><br>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;... subordinate code ...<br>
<b><tt>&nbsp;&nbsp;&nbsp;} else {</tt></b><br>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;... subordinate code ...<br>
<b><tt>&nbsp;&nbsp;&nbsp;}</tt></b><br>
<b><tt>}</tt></b>
</blockquote>

<p>
Similarly, for do-while statements, do this:

<blockquote>
<b><tt>do {</tt></b><br>
&nbsp;&nbsp;&nbsp;... body ...<br>
<b><tt>} while (</tt></b> ... condition ... <b><tt>);</tt></b>
</blockquote>

<h4><a name="Spaces"></a>Spaces</h4>

<p>
Do put a space:
<ul>
<li>after every comma and semicolon, unless it ends a line;
<li>around every binary operator other than "<b><tt>-&gt;</tt></b>" and
"<b><tt>.</tt></b>", although you can omit the spaces around the innermost
operator in a nested expression if you like;
<li>on both sides of the parentheses of an <b><tt>if</tt></b>, <b><tt>for</tt></b>, or <b><tt>while</tt></b>.
</ul>

<p>
Don't put a space:
<ul>
<li>at the end of a line;
<li>before a comma or semicolon;
<li>after unary prefix operators;
<li>before the parenthesis of a macro or procedure call.
</ul>

<h4><a name="Parentheses"></a>Parentheses</h4>

<p>
Parentheses are important in only a few places:

<ul>
<li>Around the inner subexpressions in expressions that mix
<b><tt>&amp;&amp;</tt></b> and <b><tt>||</tt></b>, even if they are not
required by precedence, for example

<blockquote><b><tt>
(xx &amp;&amp; yy) || zz
</tt></b></blockquote>

<li>Similarly around inner subexpressions in expressions that mix
<b><tt>&amp;</tt></b>, <b><tt>|</tt></b>, or shifts, especially if mixing
these with other operators, for instance

<blockquote><b><tt>
(x &lt;&lt; 3) | (y &gt;&gt; 5)
</tt></b></blockquote>

<li>In macro definitions around every use of an argument that logically
could be an expression, for example

<blockquote><b><tt>
((x) * (x) + (y) * (y))
</tt></b></blockquote>

</ul>

<p>
Anywhere else, given the choice, use fewer parentheses.

<p>
For stylistic consistency with the existing Ghostscript code, put
parentheses around conditional expressions even if they aren't
syntactically required, unless you really dislike doing this.  Note that
the parentheses should go around the entire expression, not the condition.
For instance, instead of

<blockquote><b><tt>
hpgl_add_point_to_path(pgls, arccoord_x, arccoord_y,<br>
&nbsp;&nbsp;&nbsp;(pgls-&gt;g.pen_down) ? gs_lineto : gs_moveto);
</tt></b></blockquote>

<p>
use

<blockquote><b><tt>
hpgl_add_point_to_path(pgls, arccoord_x, arccoord_y,<br>
&nbsp;&nbsp;&nbsp;(pgls-&gt;g.pen_down ? gs_lineto : gs_moveto));
</tt></b></blockquote>

<h3><a name="Preprocessor_style"></a>Preprocessor</h3>

<h4>Conditionals</h4>

<p>
Using preprocessor conditionals can easily lead to unreadable code, since
the eye really wants to read linearly rather than having to parse the
conditionals just to figure out what code is relevant.  It's OK to use
conditionals that have small scope and that don't change the structure or
logic of the program (typically, they select between different sets of
values depending on some configuration parameter), but where possible, break
up source modules rather than use conditionals that affect the structure or
logic.

<h4>Macros</h4>

<p>
Ghostscript code uses macros heavily to effectively extend the rather
weak abstraction capabilities of the C language, specifically in the area of
memory management and garbage collection: in order to read Ghostscript code
effectively, you simply have to learn some of these macros as though they
were part of the language.  The current code also uses macros heavily for
other purposes, but we are trying to phase these out as rapidly as possible,
because they make the code harder to read and debug, and to use the
rules that follow consistently in new code.

<p>
Define macros in the smallest scope you can manage (procedure, file, or
<b><tt>.h</tt></b> file), and <b><tt>#undef</tt></b> them at the end of
that scope: that way, someone reading the code can see the definitions
easily when reading the uses.  If that isn't appropriate, define them in as
large a scope as possible, so that they effectively become part of the
language.  This places an additional burden on the reader, but it can be
amortized over reading larger amounts of code.

<p>
Try hard to use procedures instead of macros.  Use "<b><tt>inline</tt></b>"
if you really think the extra speed is needed, but only within a
<b><tt>.c</tt></b> file: don't put inline procedures in <b><tt>.h</tt></b>
files, because most compilers don't honor "<b><tt>inline</tt></b>" and
you'll wind up with a copy of the procedure in every <b><tt>.c</tt></b>
file that includes the <b><tt>.h</tt></b> file.

<p>
If you define a macro that looks like a procedure, make sure it will work
wherever a procedure will work.  In particular, put parentheses around every
use of an argument within the macro body, so that the macro will parse
correctly if some of the arguments are expressions, and put parentheses
around the entire macro body.  (This is still subject to the problem that an
argument may be evaluated more than once, but there is no way around this in
C, since C doesn't provide for local variables within expressions.)

<p>
If you define macros for special loop control structures, make their uses
look somewhat like ordinary loops, for instance:

<blockquote>
<b><tt>BEGIN_RECT(xx, yy) {</tt></b><br>
&nbsp;&nbsp;... body indented one position ...<br>
<b><tt>} END_RECT(xx, yy);</tt></b>
</blockquote>

<p>
If at all possible, don't use free variables in macros -- that is, variables
that aren't apparent as arguments of the macro.  If you must use free
variables, list them all in a comment at the point where the macro is
defined.

<h3><a name="Comments"></a>Comments</h3>

<p>
The most important descriptive comments are ones in header files that
describe structures, including invariants; but every procedure or structure
declaration, or group of other declarations, should have a comment.  Don't
spend a lot of time commenting executable code unless something unusual or
subtle is going on.

<h3><a name="Naming"></a>Naming</h3>

<p>
Use fully spelled-out English words in names, rather than contractions.
This is most important for procedure and macro names, global variables and
constants, values of <b><tt>#define</tt></b> and <b><tt>enum</tt></b>,
<b><tt>struct</tt></b> and other <b><tt>typedef</tt></b> names, and
structure member names, and for argument and variable names which have
uninformative types like <b><tt>int</tt></b>.  It's not very important for
arguments or local variables of distinctive types, or for local index or
count variables.

<p>
Avoid names that run English words together:
"<b><tt>hpgl_compute_arc_center</tt></b>" is better than
"<b><tt>hpgl_compute_arccenter</tt></b>".  However, for terms drawn from
some predefined source, like the names of PostScript operators, use a term
in its well-known form (for instance, <b><tt>gs_setlinewidth</tt></b>
rather than <b><tt>gs_set_line_width</tt></b>).

<p>
Procedures, variables, and structures visible outside a single
<b><tt>.c</tt></b> file should generally have prefixes that indicate what
subsystem they belong to (in the case of Ghostscript, <b><tt>gs_</tt></b>
or <b><tt>gx_</tt></b>).  This rule isn't followed very consistently.

<h3><a name="Types"></a>Types</h3>

<p>
Many older structure names don't have <b><tt>_t</tt></b> on the end, but
this suffix should be used in all new code.  (The <b><tt>_s</tt></b>
structure name is needed only to satisfy some debuggers.  No code other than
the structure declaration should refer to it.)

<p>
Declare structure types that contain pointers to other instances of
themselves like this:

<blockquote>
<b><tt>typedef struct xxx_s xxx_t;</tt></b><br>
<b><tt>struct xxx_s {</tt></b><br>
&nbsp;&nbsp;&nbsp;... members ...<br>
&nbsp;&nbsp;&nbsp;<b><tt>xxx_t *</tt></b>ptr_member_name;<br>
&nbsp;&nbsp;&nbsp;... members ...<br>
<b><tt>};</tt></b>
</blockquote>

<p>
If, to maintain data abstraction and avoid including otherwise unnecessary
header files, you find that you want the type <b><tt>xxx_t</tt></b> to be
available in a header file that doesn't include the definition of the
structure <b><tt>xxx_s</tt></b>, use this approach:

<blockquote>
<b><tt>#ifndef xxx_DEFINED</tt></b><br>
<b><tt>#&nbsp;&nbsp;define xxx_DEFINED</tt></b><br>
<b><tt>typedef struct xxx_s xxx_t;</tt></b><br>
<b><tt>#endif</tt></b><br>
<b><tt>struct xxx_s {</tt></b><br>
&nbsp;&nbsp;&nbsp;... members ...<br>
<b><tt>};</tt></b>
</blockquote>

<p>
You can then copy the first 4 lines in other header files.  (Don't ever
include them in an executable code file.)

<p>
Don't bother using <b><tt>const</tt></b> for anything other than with
pointers as described below.  However, in those places where it is necessary
to cast a pointer of type <b><tt>const&nbsp;T&nbsp;*</tt></b> to type
<b><tt>T&nbsp;*</tt></b>, always include a comment that explains why you are
"breaking const".

<h4>Pointers</h4>

<p>
Use <b><tt>const</tt></b> for pointer referents (that is,
<b><tt>const&nbsp;T&nbsp;*</tt></b>) wherever possible and appropriate.

<p>
If you find yourself wanting to use <b><tt>void&nbsp;*</tt></b>, try to
find an alternative using unions or (in the case of super- and subclasses)
casts, unless you're writing something like a memory manager that really
treats memory as opaque.

<h3><a name="Procedures_style"></a>Procedures</h3>

<p>
In general, don't create procedures that are private and only called from
one place.  However, if a compound statement (especially an arm of a
conditional) is too long for the eye easily to match its enclosing braces
"<b><tt>{...}</tt></b>" -- that is, longer than 10 or 15 lines -- and it
doesn't use or set a lot of state held in outer local variables, it may be
more readable if you put it in a procedure.

<h3>Miscellany</h3>

<h4><a name="Local_variables"></a>Local variables</h4>

<p>
Don't assign new values to procedure parameters.  It makes debugging very
confusing when the parameter values printed for a procedure are not the
ones actually supplied by the caller.  Instead use a separate local
variable initialized to the value of the parameter.

<p>
If a local variable is only assigned a value once, assign it that value at
its declaration, if possible.  For example,

<blockquote>
<b><tt>int x = </tt></b>some expression <b><tt>;</tt></b>
</blockquote>

<p>
rather than

<blockquote>
<b><tt>int x;</tt></b><br>
...<br>
<b><tt>x = </tt></b> some expression <b><tt>;</tt></b>
</blockquote>

<p>
Use a local pointer variable like this to "narrow" pointer types:

<blockquote>
<b><tt>int</tt></b><br>
someproc(... <b><tt>gx_device *dev</tt></b> ...)<br>
<b><tt>{<br>
&nbsp;&nbsp;&nbsp;gx_device_printer *const pdev = (gx_device_printer *)dev;</tt></b><br>
&nbsp;&nbsp;&nbsp;...<br>
<b><tt>}</tt></b>
</blockquote>

<h4><a name="Compiler_warnings"></a>Compiler warnings</h4>

<p>
The following section refers to the warnings produced by <b><tt>gcc</tt></b>:
your compiler may produce different ones.

<p>
It's OK if compilation produces the following warnings:

<ul>
<li><b><tt>&lt;name&gt; might be used uninitialized in this function</tt></b>
<li><b><tt>cast discards `const' from pointer target type</tt></b>
</ul>

<p>
The first of these often occurs because the compiler isn't aware of control
flow restrictions that guarantee the variable will be initialized before
use: if it occurs in new code, check the code carefully, but don't worry
about the message.  The second is often unavoidable in code that initializes
or frees a structure that is otherwise <b><tt>const</tt></b> during its
lifetime.
<p>

Do work hard to eliminate all warnings other than these,
since they often indicate the possible presence of coding errors.
In particular, get rid of warnings about parameter passing or
initialization that discards <b><tt>const</tt></b>,
by using explicit casts.

<hr>

<h2><a name="File_structuring"></a>File structuring</h2>

<h3><a name="All_files"></a>All files</h3>

<p>
Keep file names within the "8.3" format for portability:
<ul>
<li>Use only letters, digits, dash, and underscore in file names.
<li>Don't assume upper and lower case letters are distinct.
<li>Put no more than 8 characters before the ".", if any.
<li>If there is a ".", put between 1 and 3 characters after the ".".
</ul>

<p>
For files other than documentation files, use only lower case letters
in the names; for HTML documentation files, capitalize the first letter.

<p>
Every code file should start with comments containing

<ol>
<li>a copyright notice,
<li>the name of the file in the form of an RCS Id:

<blockquote><b><tt>
/*Id$: filename.ext $*/
</tt></b></blockquote>

<p>
(using the comment convention appropriate to the language of the file), and

<li>a summary, no more than one line, of what the file contains.
</ol>

<p>
If you create a file by copying the beginning of another file, be sure to
update the copyright year and change the file name.

<h3><a name="Makefiles"></a>Makefiles</h3>

<p>
Use the extension <b><tt>.mak</tt></b> for makefiles.

<p>
For each

<blockquote><b><tt>
#include "xxx.h"
</tt></b></blockquote>

<p>
make sure there is a dependency on <b><tt>$(xxx_h)</tt></b> in the
makefile.  If xxx ends with a "<b><tt>_</tt></b>", this rule still holds,
so that if you code

<blockquote><b><tt>
#include "math_.h"
</tt></b></blockquote>

<p>
the makefile must contain a dependency on "<b><tt>$(math__h)</tt></b>"
(note the two underscores "<b><tt>__</tt></b>").

<p>
List the dependencies bottom-to-top, like the <b><tt>#include</tt></b>
statements themselves; within each level, list them alphabetically.  Do
this also with <b><tt>#include</tt></b> statements themselves whenever
possible (but sometimes there are inter-header dependencies that require
bending this rule).

<p>
For compatibility with the build utilities on OpenVMS, always put a space
before the colon that separates the target(s) of a rule from the dependents.

<h3><a name="General_C_code"></a>General C code</h3>

<p>
List <b><tt>#include</tt></b> statements from "bottom" to "top", that is,
in the following order:

<blockquote><ol>
<li>System includes (<b><tt>"xxx_.h"</tt></b>)
<li><b><tt>gs*.h</tt></b>
<li><b><tt>gx*.h</tt></b> (yes, <b><tt>gs</tt></b> and <b><tt>gx</tt></b>
are in the wrong order.)
<li><b><tt>s*.h</tt></b>
<li><b><tt>i*.h</tt></b> (or other interpreter headers that don't start
with "<b><tt>i</tt></b>")
</ol></blockquote>

<h3><a name="Headers"></a>Headers (<b><tt>.h</tt></b> files)</h3>

<p>
In header files, always use the following at the beginning of a header file
to prevent double inclusion:

<blockquote>
{{ Copyright notice etc. }}<br><br>

<b><tt>#ifndef </tt></b>&lt;filename&gt;<b><tt>_INCLUDED</tt></b><br>
<b><tt>#define </tt></b>&lt;filename&gt;<b><tt>_INCLUDED</tt></b><br><br>

{{ The contents of the file }}<br><br>

<b><tt>#endif /* </tt></b>&lt;filename&gt;<b><tt>_INCLUDED */</tt></b>
</blockquote>

<p>
The header file is the first place that a reader goes for
information about procedures, structures, constants, etc., so ensure that
every procedure and structure has a comment that says what it does.  Divide
procedures into meaningful groups set off by some distinguished form of
comment.

<h3><a name="Source"></a>Source (<b><tt>.c</tt></b> files)</h3>

<p>
After the initial comments, arrange C files in the following order:

<blockquote><ol>
<li><b><tt>#include</tt></b> statements
<li>Exported data declarations
<li>Explicit externs (if necessary)
<li>Forward declarations of procedures
<li>Private data declarations
<li>Exported procedures
<li>Private procedures
</ol></blockquote>

<p>
Be flexible about the order of the declarations if necessary to improve
readability.  Many older files don't follow this order, often without good
reason.

<hr>

<h2><a name="Conventions"></a>Ghostscript conventions</h2>

<h3><a name="Specific_names"></a>Specific names</h3>

<p>
The Ghostscript code uses certain names consistently for certain kinds of
values.  Some of the commonest and least obvious are these two:

<h4><a name="code"></a><b><tt>code</tt></b></h4>

<blockquote>
A value to be returned from a procedure:

<table cellpadding=0 cellspacing=0>
<tr valign=top>	<td align=right>&lt; 0
	<td>&nbsp;&nbsp;&nbsp;&nbsp;
	<td>An error code defined in
<a href="../src/gserrors.h">gserrors.h</a>
(or <a href="../src/errors.h">errors.h</a>)
<tr valign=top>	<td align=right>0
	<td>&nbsp;
	<td>Normal return
<tr valign=top>	<td align=right>&gt; 0
	<td>&nbsp;
	<td>A non-standard but successful return (which must be documented, preferably with the procedure's prototype)
</table>

</blockquote>

<h4><a name="status"></a><b><tt>status</tt></b></h4>

<blockquote>
A value returned from a stream procedure:

<table cellpadding=0 cellspacing=0>
<tr valign=top>	<td align=right>&lt; 0
	<td>&nbsp;&nbsp;&nbsp;&nbsp;
	<td>An exceptional condition as defined in
<a href="../src/scommon.h">scommon.h</a>
<tr valign=top>	<td align=right>0
	<td>&nbsp;
	<td>Normal return (or, from the "<b><tt>process</tt></b>" procedure, means that more input is needed)
<tr valign=top>	<td align=right>1
	<td>&nbsp;
	<td>More output space is needed (from the "<b><tt>process</tt></b>" procedure)
</table>
</blockquote>

<h3><a name="Structure_type_descriptors"></a>Structure type descriptors</h3>

<p>
The Ghostscript memory manager requires run-time type information for every
structure.  (We don't document this in detail here: see the <a
href="Develop.htm#Structure_descriptors">Structure descriptors</a> section
of the developer documentation for details.)  Putting the descriptor for a
structure next to the structure definition will help keep the two
consistent, so immediately after the definition of a structure
<b><tt>xxx_s</tt></b>, define its structure descriptor:

<blockquote>
<b><tt>struct xxx_s {</tt></b><br>
&nbsp;&nbsp;&nbsp;... members ...<br>
<b><tt>};</tt></b><br>
<b><tt>#define private_st_xxx()&nbsp;&nbsp;/* in </tt></b>&lt;filename&gt;<tt><b>.c */\</tt></b><br>
<b><tt>&nbsp;&nbsp;gs_private_st_</tt></b>&lt;whatever&gt;<b><tt>(st_xxx, xxx_t,\</tt></b><br>
<b><tt>&nbsp;&nbsp;&nbsp;&nbsp;"xxx_t", xxx_enum_ptrs, xxx_reloc_ptrs,\</tt></b><br>
<b><tt>&nbsp;&nbsp;&nbsp;&nbsp;</tt></b>... additional parameters as needed ...<b><tt>)</tt></b>
</blockquote>

<p>
The file that implements operations on this structure
(&lt;filename&gt;<b><tt>.c</tt></b>) should then include, near the
beginning, the line:

<blockquote>
<b><tt>private_st_xxx();</tt></b>
</blockquote>

<p>
In much existing code, structure descriptors are declared as
<b><tt>public</tt></b>, which allows clients to allocate instances of the
structure directly.  We now consider this bad design.  Instead, structure
descriptors should always be <b><tt>private</tt></b>; the implementation
file should provide one or more procedures for allocating instances, e.g.,

<blockquote>
<b><tt>xxx_t *gs_xxx_alloc(P1(gs_memory_t *mem));</tt></b>
</blockquote>

<p>
If it is necessary to make a structure descriptor public, it should be
declared in its clients as

<blockquote>
<b><tt>extern_st(st_xxx);</tt></b>
</blockquote>

<h3><a name="Objects"></a>"Objects"</h3>

<p>
Ghostscript makes heavy use of object-oriented constructs, including
analogues of classes, instances, subclassing, and class-associated
procedures.  However, these constructs are implemented in C rather than C++,
for two reasons:

<ul>

<li>The first Ghostscript code was written in 1986, long before C++ was
codified or was well supported by tools.  Even today, C++ tools rarely
support C++ as well as C tools support C.

<li>C++ imposes its own implementations for virtual procedures, inheritance,
run-time type information, and (to some extent) memory management.
Ghostscript requires use of its own memory manager, and also sometimes
requires the ability to change the virtual procedures of an object
dynamically.

</ul>

<h4>Classes</h4>

<p>
The source code representation of a class is simply a
<b><tt>typedef</tt></b> for a C <b><tt>struct</tt></b>.  See <a
href="C-style.htm#Structures">Structures</a>, above, for details.

<h4>Procedures</h4>

<p>
Ghostscript has no special construct for non-virtual procedures associated
with a class.  In some cases, the <b><tt>typedef</tt></b> for the class is
in a header file but the <b><tt>struct</tt></b> declaration is in the
implementation code file: this provides an extra level of opaqueness, since
clients then can't see the representation and must make all accesses through
procedures.  You should use this approach in new code, if it doesn't
increase the size of the code too much or require procedure calls for very
heavily used accesses.

<p>
Ghostscript uses three different approaches for storing and accessing
virtual procedures, plus a fourth one that is recommended but not currently
used.  For exposition, we assume the class (type) is named
<b><tt>xxx_t</tt></b>, it has a virtual procedure
<b><tt>void&nbsp;(*virtu)(P1(xxx_t&nbsp;*))</tt></b>, and we have a variable
declared as <b><tt>xxx_t&nbsp;*pxx</tt></b>.

<ol>

<li>The procedures are stored in a separate, constant structure of type
<b><tt>xxx_procs</tt></b>, of which <b><tt>virtu</tt></b> is a member.  The
structure definition of <b><tt>xxx_t</tt></b> includes a member defined as
<b><tt>const&nbsp;xxx_procs&nbsp;*procs</tt></b> (always named
<b><tt>procs</tt></b>).  The construct for calling the virtual procedure is
<b><tt>pxx->procs->virtu(pxx)</tt></b>.

<li>The procedures are defined in a structure of type
<b><tt>xxx_procs</tt></b> as above.  The structure definition of
<b><tt>xxx_t</tt></b> includes a member defined as
<b><tt>xxx_procs&nbsp;procs</tt></b> (always named <b><tt>procs</tt></b>).
The construct for calling the virtual procedure is
<b><tt>pxx->procs.virtu(pxx)</tt></b>.

<li>The procedures are not defined in a separate structure: each procedure
is a separate member of <b><tt>xxx_t</tt></b>.  The construct for calling
the virtual procedure is <b><tt>pxx->virtu(pxx)</tt></b>.

<li>The procedures are defined in a structure of type
<b><tt>xxx_procs</tt></b> as above.  The structure definition of
<b><tt>xxx_t</tt></b> includes a member defined as
<b><tt>xxx_procs&nbsp;procs[1]</tt></b> (always named
<b><tt>procs</tt></b>).  The construct for calling the virtual procedure is
again <b><tt>pxx->procs->virtu(pxx)</tt></b>.

</ol>

<p>
Note that in approach 1, the procedures are in a shared constant structure;
in approaches 2 - 4, they are in a per-instance structure that can be
changed dynamically, which is sometimes important.

<p>
In the present Ghostscript code, approach 1 is most common, followed by 2
and 3; 4 is not used at all.  For new code, you should use 1 or 4: that way,
all virtual procedure calls have the same form, regardless of whether the
procedures are shared and constant or per-instance and mutable.

<h4>Subclassing</h4>

<p>
Ghostscript's class mechanism allows for subclasses that can add data
members, or can add procedure members if approach 1 or 3 (above) is used.
Since C doesn't support subclassing, we use a convention to accomplish it.
In the example below, <b><tt>gx_device</tt></b> is the root class; it has a
subclass <b><tt>gx_device_forward</tt></b>, which in turn has a subclass
<b><tt>gx_device_null</tt></b>.  First we define a macro for all the members
of the root class, and the root class type.  (As for structures in general,
classes need a structure descriptor, as discussed in <a
href="#Structures">Structures</a> above: we include these in the examples
below.)

<blockquote><b><tt>
#define gx_device_common\<br>
&nbsp;&nbsp;&nbsp;&nbsp;type1 member1;\<br>
&nbsp;&nbsp;&nbsp;&nbsp;</tt></b>...<b><tt><br>
&nbsp;&nbsp;&nbsp;&nbsp;typeN memberN<br>
<br>
typedef struct gx_device_s {<br>
&nbsp;&nbsp;&nbsp;&nbsp;gx_device_common;<br>
} gx_device;<br>
<br>
#define private_st_gx_device()&nbsp;&nbsp;/* in gsdevice.c */\<br>
&nbsp;&nbsp;gs_private_st_</tt></b>&lt;whatever&gt;<b><tt>(st_gx_device, gx_device,\<br>
&nbsp;&nbsp;&nbsp;&nbsp;"gx_device", device_enum_ptrs, device_reloc_ptrs,\<br>
&nbsp;&nbsp;&nbsp;&nbsp;</tt></b>... additional parameters as needed ...<b><tt>)</tt></b>
</tt></b></blockquote>

<p>
We then define a similar macro and type for the subclass.

<blockquote><b><tt>
#define gx_device_forward_common\<br>
&nbsp;&nbsp;&nbsp;&nbsp;gx_device_common;\<br>
&nbsp;&nbsp;&nbsp;&nbsp;gx_device *target<br>
<br>
typedef struct gx_device_forward_s {<br>
&nbsp;&nbsp;&nbsp;&nbsp;gx_device_forward_common;<br>
} gx_device_forward;<br>
<br>
#define private_st_device_forward()&nbsp;&nbsp;/* in gsdevice.c */\<br>
&nbsp;&nbsp;gs_private_st_suffix_add1(st_device_forward, gx_device_forward,\<br>
&nbsp;&nbsp;&nbsp;&nbsp;"gx_device_forward", device_forward_enum_ptrs, device_forward_reloc_ptrs,\<br>
&nbsp;&nbsp;&nbsp;&nbsp;gx_device, target)
</tt></b></blockquote>

<p>
Finally, we define a leaf class, which doesn't need a macro because we don't
currently subclass it.  (We can create the macro later if needed, with no
changes anywhere else.)  In this particular case, the leaf class has no
additional data members, but it could have some.

<blockquote><b><tt>
typedef struct gx_device_null_s {<br>
&nbsp;&nbsp;&nbsp;&nbsp;gx_device_forward_common;<br>
};<br>
<br>
#define private_st_device_null()&nbsp;&nbsp;/* in gsdevice.c */\<br>
&nbsp;&nbsp;gs_private_st_suffix_add0_local(st_device_null, gx_device_null,\<br>
&nbsp;&nbsp;&nbsp;&nbsp;"gx_device_null", device_null_enum_ptrs, device_null_reloc_ptrs,\<br>
&nbsp;&nbsp;&nbsp;&nbsp;gx_device_forward)
</tt></b></blockquote>

<p>
Note that the above example is <strong>not</strong> the actual definition of
the <b><tt>gx_device</tt></b> structure type: the actual type has some
additional complications because it has a finalization procedure.  See <a
href="../src/gxdevcli.h">src/gxdevcli.h</a> for the details.

<p>
If you add members to a root class (such as <b><tt>gx_device</tt></b> in
this example), or change existing members, do this in the
<b><tt>gx_device_common</tt></b> macro, not the <b><tt>gx_device</tt></b>
structure definition.  Similarly, to change the
<b><tt>gx_device_forward</tt></b> class, modify the
<b><tt>gx_device_forward_common</tt></b> macro, not the structure
definition.  Only change the structure definition if the class is a leaf
class (one with no <b><tt>_common</tt></b> macro and no possibility of
subclassing), like <b><tt>gx_device_null</tt></b>.

<h3><a name="Error_handling"></a>Error handling</h3>

<p>
Every caller should check for error returns and, in general, propagate them
to <b>its</b> callers.  By convention, nearly every procedure returns an
<b><tt>int</tt></b> to indicate the outcome of the call:

<blockquote><table cellpadding=0 cellspacing=0>
<tr valign=top>	<td align=right>&lt; 0
	<td>&nbsp;&nbsp;&nbsp;&nbsp;
	<td>Error return
<tr valign=top>	<td align=right>0
	<td>&nbsp;
	<td>Normal return
<tr valign=top>	<td align=right>&gt; 0
	<td>&nbsp;
	<td>Non-error return other than the normal case
</table></blockquote>

<p>
To make a procedure generate an error and return it, as opposed to
propagating an error generated by a lower procedure, you should use

<blockquote>
<b><tt>return_error(</tt></b><em>error_number</em><b><tt>);</tt></b>
</blockquote>

<p>
Sometimes it is more convenient to generate the error in one place and
return it in another.  In this case, you should use

<blockquote>
<b><tt>code = gs_note_error(</tt></b><em>error_number</em><b><tt>);</tt></b><br>
...<br>
<b><tt>return code;</tt></b>
</blockquote>

<p>
In executables built for debugging, the <b><tt>-E</tt></b> (or
<b><tt>-Z#</tt></b>) command line switch causes <b><tt>return_error</tt></b>
and <b><tt>gs_note_error</tt></b> to print the error number and the source
file and line: this is often helpful for identifying the original cause of
an error.

<p>
See the file <a href="../src/gserrors.h">src/gserrors.h</a> for the error
return codes used by the graphics library, most of which correspond directly
to PostScript error conditions.

<!-- [2.0 end contents] ==================================================== -->

<!-- [3.0 begin visible trailer] =========================================== -->
<hr>

<p>
<small>Copyright &copy; 1996, 1997, 1998 Aladdin Enterprises.
Copyright &copy; 2001 artofcode LLC.
All rights reserved.</small>

<p>
<small>This file is part of AFPL Ghostscript.  See the <a
href="Public.htm">Aladdin Free Public License</a> (the "License") for full
details of the terms of using, copying, modifying, and redistributing AFPL
Ghostscript.</small>

<p>
<small>Ghostscript version 7.04, 31 January 2002

<!-- [3.0 end visible trailer] ============================================= -->

</body>
</html>