• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"
2          "http://www.w3.org/TR/html4/strict.dtd">
3<html>
4<head>
5  <META http-equiv="Content-Type" content="text/html; charset=ISO-8859-1">
6  <title>Clang - Get Involved</title>
7  <link type="text/css" rel="stylesheet" href="menu.css">
8  <link type="text/css" rel="stylesheet" href="content.css">
9</head>
10<body>
11
12<!--#include virtual="menu.html.incl"-->
13
14<div id="content">
15
16<h1>Open Clang Projects</h1>
17
18<p>Here are a few tasks that are available for newcomers to work on, depending
19on what your interests are.  This list is provided to generate ideas, it is not
20intended to be comprehensive.  Please ask on cfe-dev for more specifics or to
21verify that one of these isn't already completed. :)</p>
22
23<ul>
24<li><b>Undefined behavior checking</b>:
25Improve and extend the runtime checks for undefined behavior which CodeGen
26inserts for the various <tt>-fsanitize=</tt> modes. A lot of issues can already
27be caught, but there is more to do here.</li>
28
29<li><b>Improve target support</b>: The current target interfaces are heavily
30stubbed out and need to be implemented fully.  See the FIXME's in TargetInfo.
31Additionally, the actual target implementations (instances of TargetInfoImpl)
32also need to be completed.</li>
33
34<li><b>Implement an tool to generate code documentation</b>: Clang's
35library-based design allows it to be used by a variety of tools that reason
36about source code. One great application of Clang would be to build an
37auto-documentation system like doxygen that generates code documentation from
38source code. The advantage of using Clang for such a tool is that the tool would
39use the same preprocessor/parser/ASTs as the compiler itself, giving it a very
40rich understanding of the code. Clang is already able to read and understand
41doxygen markup, but cannot yet generate documentation from it.</li>
42
43<li><b>Use clang libraries to implement better versions of existing tools</b>:
44Clang is built as a set of libraries, which means that it is possible to
45implement capabilities similar to other source language tools, improving them
46in various ways.  Three examples are <a
47href="http://distcc.samba.org/">distcc</a>, the <a
48href="http://delta.tigris.org/">delta testcase reduction tool</a>, and the
49"indent" source reformatting tool.
50distcc can be improved to scale better and be more efficient.  Delta could be
51faster and more efficient at reducing C-family programs if built on the clang
52preprocessor. The clang-based indent replacement,
53<a href="http://clang.llvm.org/docs/ClangFormat.html">clang-format</a>,
54could be taught to handle simple structural rules like those in <a
55href="http://llvm.org/docs/CodingStandards.html#hl_earlyexit">the LLVM coding
56standards</a>.</li>
57
58<li><b>Use clang libraries to extend Ragel with a JIT</b>: <a
59href="http://research.cs.queensu.ca/~thurston/ragel/">Ragel</a> is a state
60machine compiler that lets you embed C code into state machines and generate
61C code.  It would be relatively easy to turn this into a JIT compiler using
62LLVM.</li>
63
64<li><b>Self-testing using clang</b>: There are several neat ways to
65improve the quality of clang by self-testing. Some examples:
66<ul>
67  <li>Improve the reliability of AST printing and serialization by
68  ensuring that the AST produced by clang on an input doesn't change
69  when it is reparsed or unserialized.
70
71  <li>Improve parser reliability and error generation by automatically
72  or randomly changing the input checking that clang doesn't crash and
73  that it doesn't generate excessive errors for small input
74  changes. Manipulating the input at both the text and token levels is
75  likely to produce interesting test cases.
76</ul>
77</li>
78
79<li><b>Continue work on C++1y support</b>:
80  C++98 and C++11 are feature-complete, but there are still several C++1y features to
81  implement.  Please see the <a href="cxx_status.html">C++ status report
82  page</a> to find out what is missing.</li>
83
84<li><b>StringRef'ize APIs</b>: A thankless but incredibly useful project is
85StringRef'izing (converting to use <tt>llvm::StringRef</tt> instead of <tt>const
86char *</tt> or <tt>std::string</tt>) various clang interfaces. This generally
87simplifies the code and makes it more efficient.</li>
88
89<li><b>Universal Driver</b>: Clang is inherently a cross compiler. We would like
90to define a new model for cross compilation which provides a great user
91experience -- it should be easy to cross compile applications, install support
92for new architectures, access different compilers and tools, and be consistent
93across different platforms. See the <a href="UniversalDriver.html">Universal
94Driver</a> web page for more information.</li>
95
96<li><b>XML Representation of ASTs</b>: Clang maintains a rich Abstract Syntax Tree that describes the program. Clang could emit an XML document that describes the program, which others tools could consume rather than being tied directly to the Clang binary.The XML representation needs to meet several requirements:
97  <ul>
98    <li><i>General</i>, so that it's able to represent C/C++/Objective-C abstractly, and isn't tied to the specific internal ASTs that Clang uses.</li>
99    <li><i>Documented</i>, with appropriate Schema against which the output of Clang's XML formatter can be verified.</li>
100    <li><i>Stable</i> across Clang versions.</li>
101  </ul></li>
102
103<li><b>Configuration Manager</b>: Clang/LLVM works on a large number of
104architectures and operating systems and can cross-compile to a similarly large
105number of configurations, but the pitfalls of chosing the command-line
106options, making sure the right sub-architecture is chosen and that the correct
107optional elements of your particular system can be a pain.
108
109<p>A tool that would investigate hosts and targets, and store the configuration
110in files that can later be used by Clang itself to avoid command-line options,
111especially the ones regarding which target options to use, would greatle alleviate
112this problem. A simple tool, with little or no dependency on LLVM itself, that
113will investigate a target architecture by probing hardware, software, libraries
114and compiling and executing code to identify all properties that would be relevant
115to command-line options (VFP, SSE, NEON, ARM vs. Thumb etc), triple settings etc.</p>
116
117<p>The first stage is to build a CFLAGS for Clang that would produce code on the
118current Host to the identified Target.</p>
119
120<p>The second stage would be to produce a configuration file (that can be used
121independently of the Host) so that Clang can read it and not need a gazillion
122of command-line options. Such file should be simple JSON / INI or anything that
123a text editor could change.</p>
124</ul>
125
126<p>If you hit a bug with clang, it is very useful for us if you reduce the code
127that demonstrates the problem down to something small.  There are many ways to
128do this; ask on cfe-dev for advice.</p>
129
130</div>
131</body>
132</html>
133