Sophie

Sophie

distrib > PLD > ac > amd64 > media > dist > by-pkgid > 14f9a678898ad5b0a0a20e99547b9081 > files > 5

retawq-0.2.6b-1.amd64.rpm

<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd">
<html version="4.01"><head><title>[retawq] Coding Style</title></head>
<body text="#000000" bgcolor="#cccccc" link="#0000ff" vlink="#551a8b"
alink="#551a8b" lang="en">
<center><b><font size="+2">retawq Documentation</font><br><font
size="+1">Coding Style</font></b></center>

<p><a name="id"></a><b>Identifiers</b></p>

<p><b>Type</b> identifiers consist of a lowercase "t" and meaningful, mixedcase
notions. <b>Variable</b> identifiers are normally written in lowercase letters,
their parts (if any) are separated by underscore ("_") characters.
<b>Constant</b> identifiers often consist of a lowercase prefix formed by
uppercase letters from the corresponding type identifier (if that's
non-trivial), followed by a mixedcase, meaningful notion; a few constant
identifiers are instead implemented using preprocessor defines, especially
concerning "hard limits" for variables and some constants the values of which
have to be accessible by the preprocessor for build-time error checking.
<b>Function</b> identifiers are written like variable identifiers;
additionally, their first part often refers to the source code file (or
conceptual project part) they belong to.</p>

<p><a name="type"></a><b>Using Types</b></p>

<p>If a type is needed in several parts of the program, we define a new type
identifier using "typedef". This way, we only have to change one place when we
see that the original choice for that type wasn't "optimal".</p>

<p>Enumeration types are <i>not</i> defined with "typedef enum"; variables of
such a type would waste much memory because they're "int"s. Instead, we define
the constants using "enum" and define the type itself using "typedef unsigned
char ...." or whatever might be appropriate. Additionally, we don't rely on the
compiler assigning certain numbers to enumeration identifiers, instead we
assign the numbers ourselves.</p>

<p>We don't rely on the "fact" that e.g. a <i>char</i> declaration is always
interpreted as <i>unsigned char</i> by the compiler - some compilers interpret
it as <i>signed char</i>. So we explicitly say <i>signed char</i> resp.
<i>unsigned char</i> (unless we're actually using the <i>char</i> type for
characters, and of course except for cases where e.g. a curses library header
file simply says "short"...).</p>

<p><a name="mark"></a><b>Marks</b></p>

<p>The source code contains lots of comments. Some of them are tagged with
special "marks", so that their purpose becomes obvious at first glance and they
can easily be recovered with "search" or "grep" commands. The following marks
are used:</p>

<p><ul>
<li><b>NOTE:</b> a comment which is important for understanding the source
code</li>
<li><b>IMPORTANT:</b> a non-obvious fact which is important to know of for
getting the program behavior right</li>
<li><b>FIXME:</b> a known bug or unintended program behavior; public releases
shouldn't contain such marks, but there might be some "unimportant" bugs which
shouldn't delay a release...</li>
<li><b>CHECKME:</b> code which should be revised or re-checked some time; this
mostly concerns program design decisions which seem to be not satisfying or
have some flaws.</li>
<li><b>IMPLEMENTME:</b> a feature idea which might be interesting and should be
implemented some time</li>
<li><b>IMPROVEME:</b> code which is known to run slower than it could, but
often the performance improvement will make the code much more complicated;
since retawq is already pretty fast, such little performance problems aren't
very interesting...</li>
<li><b>REMOVEME:</b> commented-out code which has been replaced by better code
or isn't needed any longer for other reasons, e.g. old debugging messages;
should be removed some time</li>
<li><b>detach:</b> we didn't forget to deallocate memory, we intentionally only
detached a pointer.</li>
<li><b>"should not happen":</b> conditions which won't be true during normal
program execution but might become true due to bugs; such code is only there in
order to prevent the program from crashing (or other bad behavior) in case of
bugs.</li>
<li><b>"can't happen":</b> conditions which are much less likely than the
"should not happen" stuff; probably paranoid code</li>
<li><b>#if 0:</b> code which has been disabled because it isn't needed any
longer or doesn't yet work as it should or requires changes somewhere else
before it can work</li>
</ul></p>

<p>Now if you still don't understand the source code, I give up... :-)</p>

<p><hr>This documentation file is part of version 0.2.6b of <a
href="http://retawq.sourceforge.net/">retawq</a>, a network client created by
<span lang="de">Arne Thoma&szlig;en</span>. retawq is basically released under
certain versions of the GNU General Public License and WITHOUT ANY WARRANTY.
Copyright (C) 2001-2005 <a href="mailto:arne@arne-thomassen.de"><span
lang="de">Arne Thoma&szlig;en</span></a>.</p>
</body></html>