1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
|
<html><head><title>toybox cleanup</title></head>
<!--#include file="header.html" -->
<h1>Index</h1>
<ul>
<li><a href=#intro>Introduction</a></li>
<li><a href=#advice>Advice</a></li>
<li>Commands:</li>
<ul>
<li><a href=#uuencode>uuencode</a></li>
<li><a href=#uudecode>uudecode</a></li>
<li><a href=#ifconfig>ifconfig</a></li>
</ul>
</ul>
<hr>
<a name=intro />
<h1>Cleaning up the toybox code.</h1>
<p>Toybox <a href=http://landley.net/notes.html#31-03-2013>hasn't always</a>
taken proper advantage of external contributions</a>.
Lots of people want to help, but their contributions languish out of tree
or in the "pending" directory, awaiting cleanup.</p>
<p>Toybox's design goals require simpler, tighter, and more explicit code
than most other implementations, among other reasons to allow better security
auditing. Writing "another" implementation of standard command line tools
isn't very interesting, they should be _better_ implementations.
Unfortunately, this means existing, working commands often take more effort to
clean up to Toybox's standards than writing a new one from scratch, not
because they don't work, but because we aim for an unusual level of polish.</p>
<p>In hopes of teaching more people how to do this
cleanup work, I've started breaking cleanup changes into smaller chunks and
posting explanations of each change to the mailing list.
Below are indexes of such cleanup series. Each commit and post are meant to
be read together: each description explains what the corresponding patch
was trying to accomplish.</p>
<p>Line/byte totals of completed series are given for scale, but the point
of this work is simplicity and compactness, not size per se.</p>
<p>(Note: mercurial's web viewer doesn't follow renames, so although each
command name links to a commit list with the bulk of the changes, it may
not include the final version of each file moved from the "pending"
directory to its final location. The summaries link the initial and cleaned
versions of each file when giving line counts.)</p>
<hr>
<a name=advice />
<h1>General advice and/or policy.</h1>
<p>The <a href=design.html>design of toybox</a> page and the coding
style section at the start of the <a href=code.html>source code walkthrough</a>
don't cover everything. Here are some
links to mailing list messages that cover various programming topics
not directly related to a specific cleanup series:</p>
<ul>
<li><a href=http://lists.landley.net/pipermail/toybox-landley.net/2013-April/000850.html>Error messages and internationalization.</a></li>
<li><a href=http://lists.landley.net/pipermail/toybox-landley.net/2013-April/000891.html>Why not "const"?</a> (Exception: global variables
outside of GLOBALS can be static const, to go in rodata instead of data.
This means the pages can be shared between instances.)</li>
<li><a href=http://lkml.indiana.edu/hypermail/linux/kernel/1308.3/03890.html>Why not "bool"?</a> (explanation from Linus Torvalds)</li>
<li><a href=http://lists.landley.net/pipermail/toybox-landley.net/2013-April/000893.html>Why not to check in debug code.</a></li>
<li><a href=http://lists.landley.net/pipermail/toybox-landley.net/2013-June/001044.html>Relationship between /proc and /sys</a> (/proc isn't obsolete and /sys is an ABI)</li>
<li>"Hiding numbers that are used just once into defines to put them out of
sight does not really help readability."</a> -
<a href=http://lkml.iu.edu/hypermail/linux/kernel/1407.1/00299.html>Pavel
Machek</a></li>
<li>"Infrastructure in search of a user" is a bad thing, so don't put code
in lib/ that doesn't already have at least two users. Don't preemptively
factor stuff out, it's easy enough to rewrite it uin future if it needs
to change. The "extreme programming" fad called this "You Ain't Gonna
Need It" (while inexplicably screaming at cans of Mountain Dew, back in the
90's). Here's <a href=https://lwn.net/Articles/683745/>Linus Torvalds' take</a>.</li>
</ul>
<hr>
<a name="uuencode"><h1><a href=/hg/toybox/log/900/toys/pending/uuencode.c>uuencode</a></h1>
<p>This is an example of cleaning up something most projects would be quite
happy with. The initial submission of uuencode and uudecode was high
quality code, written by a seasoned developer who did an excellent
job, but it was still possible to shrink the result almost by half:</p>
<ul>
<li>old total: <a href=/hg/toybox/file/828/toys/pending/uuencode.c>116 lines (2743
bytes) in 7 functions</a></li>
<li>new total: <a href=/hg/toybox/file/841/toys/posix/uuencode.c>67 lines (1440
bytes) in 1 function</a></li>
</ul>
<ul>
<li>commit: <a href=/hg/toybox/rev/830>830</a>: first pass, description: <a href=http://lists.landley.net/pipermail/toybox-landley.net/2013-April/000904.html>part 1</a>,
<a href=http://lists.landley.net/pipermail/toybox-landley.net/2013-April/000903.html>part 2</a></li>
<li>commit: <a href=/hg/toybox/rev/831>831</a>,
second pass, description: <a href=http://lists.landley.net/pipermail/toybox-landley.net/2013-April/000919.html>part 3</a></li>
<li>commit: <a href=/hg/toybox/rev/837>837</a>,
description: fix test suite.</li>
<li>commit: <a href=/hg/toybox/rev/853>853</a>, description: bugfix.</li>
</ul>
<p>Status: COMPLETE</p>
<a name="uudecode"><h1><a href=/hg/toybox/log/900/toys/pending/uudecode.c>uudecode</a></h1>
<p>The uudecode cleanup was my second "explain as I go along" cleanup,
and I tried to do it in smaller stages so it was easier to see what
changed from the diff:</p>
<ul>
<li>old: <a href=/hg/toybox/file/828/toys/pending/uudecode.c>175
lines (4534 bytes) in 9 functions</a></li>
<li>new: <a href=/hg/toybox/file/840/toys/posix/uudecode.c>107 lines
(2300 bytes) in 1 function</a></li>
</ul>
<ul>
<li>commit: <a href=/hg/toybox/rev/833>833</a>,
description: preparatory adjustments to test suite.</li>
<li>commit: <a href=/hg/toybox/rev/835>835</a>,
description: <a href=http://lists.landley.net/pipermail/toybox-landley.net/2014-January/001532.html>Redo command line parsing, redo parsing loop.</a></li>
<li>commit: <a href=/hg/toybox/rev/838>838</a>,
description: <a href=http://lists.landley.net/pipermail/toybox-landley.net/2014-January/001533.html>Redo b64_1byte, b64_4bytes, and uu_line()</a></li>
<li>commit: <a href=/hg/toybox/rev/839>839</a>,
description: todo</a></li>
<li>commit: <a href=/hg/toybox/rev/840>840</a>,
description: todo (finish, move pending->posix, default y)</a></li>
</ul>
<p>Status: COMPLETE</p>
<a name=ifconfig>
<h1><a href=/hg/toybox/log/tip/toys/pending/ifconfig.c>ifconfig</a></h1>
<p>This series describes a thorough cleanup that took a while to do.</p>
<p>When ifconfig was submitted, it touched a half-dozen files. I glued it
together into a single self-contained file, which needed a lot of
cleanup. The final version is about 1/3 the size of the original.</p>
<ul>
<li>old total: <a href=/hg/toybox/file/841/toys/pending/ifconfig.c>1504 lines (44268 bytes) in 38 functions</a></li>
<li>new total: <a href=/hg/toybox/file/1133/toys/other/ifconfig.c>521 lines (15963 bytes) in 4 functions</a></li>
</ul>
<p>This was the first command I started cleaning up with the intent of
describing the process, and especially the first few entries in this
series describe many of the low hanging fruit techniques used to clean
up a codebase.</p>
<ul>
<li>commit: <a href=/hg/toybox/rev/843>843</a>, description: <a href=http://lists.landley.net/pipermail/toybox-landley.net/2013-April/000882.html>Infrastructure in search of a user, code proximity,
unnecessary #defines, shorter code is easier to read.</a></li>
<li>commit: <a href=/hg/toybox/rev/844>844</a>,
description: <a href=http://lists.landley.net/pipermail/toybox-landley.net/2013-April/000881.html>Headers, replacing repeated code with loops,
logical operators guaranteed to return 0 or 1, math on string constants,
removing unnecessary variables and tests.</a></li>
<li>commit: <a href=/hg/toybox/rev/852>852</a>,
description: <a href=http://lists.landley.net/pipermail/toybox-landley.net/2013-April/000921.html>hg commit numbers, documenting the obvious, ordering
to avoid prototypes, returning void, collate local declarations,
use error_exit(), unnecessary parentheses, inline to remove variables/functions
used only once, using *var instead of var[0], unnecessary typecasts,
xprintf("\n") vs xputc('\n')</a></li>
<li>commit: <a href=/hg/toybox/rev/856>856</a>,
description: one line portability fix from Isaac Dunham</li>
<li>commit: <a href=/hg/toybox/rev/861>861</a>
and <a href=/hg/toybox/rev/863>863</a>,
description:
<a href=http://lists.landley.net/pipermail/toybox-landley.net/2013-April/000910.html>Help
infrastructure cleanup from Isaac Dunham</a>
(which I mis-applied and then <a href=http://lists.landley.net/pipermail/toybox-landley.net/2013-April/000920.html>fixed plus some whitespace changes</a>)</li>
<li>commit: <a href=/hg/toybox/rev/862>862</a>, description:
<a href=http://lists.landley.net/pipermail/toybox-landley.net/2014-January/001525.html>remove unused headers and function, replace local buffer with toybuf, perror_exit(), avoid unnecessary assignment.</a></li>
<li>commit: <a href=/hg/toybox/rev/864>864</a>, description:
<a href=http://lists.landley.net/pipermail/toybox-landley.net/2014-January/001526.html>use common linked list functions, inline set_data, add xioctl, clean up error messages, whitespace and comment tweaks, remove NOP return statements</a></li>
<li>commit: <a href=/hg/toybox/rev/866>866</a>, description:
<a href=http://lists.landley.net/pipermail/toybox-landley.net/2014-January/001528.html>move standalone globals into GLOBALS() block, collate structs into
iface_list. Inline/rewrite/remove field_format, iface_flags_str,
omit_whitespace(), print_iface_flags(), print_media(), get_ifconfig_info(),
and clear_list(). Merge duplicate function
calls. Show get_proc_info() version field can't matter in 2.6 or newer kernel,
and that SIOCGIFMAP has been there since 1994 so doesn't need an #ifdef.
Loop simplification in readconf() and show_iface().</a></li>
<li>commit: <a href=/hg/toybox/rev/869>869</a> and <a href=/hg/toybox/rev/870>870</a>,
description: <a href=http://lists.landley.net/pipermail/toybox-landley.net/2013-April/000928.html>869:
reorder to eliminate prototypes, put command_main() at end of file,
870: long repeated variable prefixes, replacing struct+sscanf()+printf with a
loop and a table (from iface_list, get_proc_info(), display_ifconfig()),
use lib/xwrap.c functions to return void, why xstrcpy() fails closed,
(functional comment: why multicast failed, CSLIP obsolecense), not being
_too_ clever.</a></li>
<li>commit: <a href=/hg/toybox/rev/878>878</a> and <a href=/hg/toybox/rev/879>879</a>:
description: <a href=http://lists.landley.net/pipermail/toybox-landley.net/2013-April/000946.html>878: add xsocket(), free(NULL) is a safe NOP (posix!).
879: inline three functions and simplify, some simplifications only show up
after repeated inlining</a></li>
<li>commit: <a href=/hg/toybox/rev/883>883</a>,
description: move some common code to lib/ and posix headers to toys.h.</li>
<li>commit: <a href=/hg/toybox/rev/898>898</a>,
description: <a href=http://lists.landley.net/pipermail/toybox-landley.net/2013-May/000974.html>Argument parsing. (Replace ifconfig_main() if/else staircase with a loop over
an array, genericize minus prefix logic, inline a use of set_flags().)</a></li>
<li>commit: <a href=/hg/toybox/rev/905>905</a>,
description: <a href=http://lists.landley.net/pipermail/toybox-landley.net/2013-May/000992.html>remove unnecessary wrapper function, inlining more functions,
relying on the values of constants that don't change across architectures
(binary backwards compatability), more ifconfig_main table work,
man ioctl_list.</a></li>
<li>commit: <a href=/hg/toybox/rev/906>906</a>,
description: <a href=http://lists.landley.net/pipermail/toybox-landley.net/2013-May/000994.html>More ifconfig_main() table work, remove vestigial arguments
to functions, "a frenzy of inlining", slightly better error reporting,
don't reinvent libc functions.</a></li>
<li>commit: <a href=/hg/toybox/rev/907>907</a>,
<a href=http://lists.landley.net/pipermail/toybox-landley.net/2013-May/000995.html>inlining show_ip_addr() with a loop and a table, inlining hex_to_binary()
and set_hw_address(), stop validating data we supplied, remove a table
that's overkill (two entries, we don't even loop over it), when you don't need a
NULL terminator for array, remove unnecessary memcpy(offsetof()) with
assignment, trusting -funsigned-char.</a></li>
<li>commit: <a href=/hg/toybox/rev/919>919</a>,
<a href=http://lists.landley.net/pipermail/toybox-landley.net/2013-June/001027.html>todo whitespace damage, introduce IFREQ_OFFSZ() and poke() to
ifconfig_main() table logic to fold more if/else parts into the table</a></li>
<li>Status update: <a href=http://lists.landley.net/pipermail/toybox-landley.net/2013-June/001033.html>Entering the home stretch on ifconfig</a> (and a <a href=http://lists.landley.net/pipermail/toybox-landley.net/2013-June/001043.html>note about infiniband</a>)</li>
<li>commit: <a href=/hg/toybox/rev/921>921</a>, description: <a href=http://lists.landley.net/pipermail/toybox-landley.net/2014-June/003508.html>Inline a couple more functions and make sockfd a global.</li>
<li>commit: <a href=/hg/toybox/rev/957>957</a>, description: <a href=http://lists.landley.net/pipermail/toybox-landley.net/2013-July/001121.html>Remove unused socklen and addr_to_len(), cleanup so we can merge get_device_info()/display_ifconfig().</a></li>
<li>commit: <a href=/hg/toybox/rev/958>958</a>, description: <a href=http://lists.landley.net/pipermail/toybox-landley.net/2013-July/001131.html>This commit removes struct if_list by unifying get_device_info() and display_ifconfig().</a></li>
<li>commit: <a href=/hg/toybox/rev/1104>1104</a>, description: Merge toynet into toys.h: musl supports it and micromanaging uClibc config options isn't very interesting anymore.</li>
<li>commit: <a href=/hg/toybox/rev/1127>1127</a>, description: <a href=http://lists.landley.net/pipermail/toybox-landley.net/2013-November/001464.html>Start tacling ipv6 issues, beginning with display_ifconfig().</a></li>
<li>commit: <a href=/hg/toybox/rev/1128>1128</a>, description: <a href=http://lists.landley.net/pipermail/toybox-landley.net/2013-November/001463.html>More ipv6, make struct sockaddr_with_len go away, merge more arguments into the table in main().</a></li>
<li>commit: <a href=/hg/toybox/rev/1132>1132</a>, description: promotion from pending to other</li>
<li>commit: <a href=/hg/toybox/rev/1133>1133</a>, description: <a href=http://lists.landley.net/pipermail/toybox-landley.net/2013-November/001462.html>cleanup help text, remove obsolete/NOP commands</a></li>
</ul>
<p>Status: COMPLETE.</p>
<p>Status: in progress.</p>
<h1><a href=/hg/toybox/log/917/toys/pending/stat.c>stat</a></h1>
<p>A lot of the stat cleanup was done by Felix Janda.</p>
<ul>
<li>commit <a href=/hg/toybox/rev/747>747</a>: initial submission</a></li>
<li>commit <a href=/hg/toybox/rev/810>810</a>: whitespace</li>
<li>commit <a href=/hg/toybox/rev/811>811</a>: description in commit message.</li>
<li>commit <a href=/hg/toybox/rev/871>871</a>: whitespace (reindent from 4 spaces to 2)</li>
<li>commit <a href=/hg/toybox/rev/872>872</a>: Felix Janda - <a href=http://lists.landley.net/pipermail/toybox-landley.net/2013-April/000923.html>cleanup</a> (with discussion thread)</li>
<li>commit <a href=/hg/toybox/rev/875>885</a>: Felix Janda - <a href=http://lists.landley.net/pipermail/toybox-landley.net/2013-April/000936.html>move permission formatting (777 -> -rwxrwxrwx) from ls to lib so stat can reuse it.</a></li>
<li>commit <a href=/hg/toybox/rev/885>886</a>: Felix Janda - remove unimplemented options and clean up help text</li>
<li>commit <a href=/hg/toybox/rev/910>910</a>: Felix Janda - <a href=http://lists.landley.net/pipermail/toybox-landley.net/2013-May/001013.html>Add support for stating multiple files</a>.</li>
<li>commit <a href=/hg/toybox/rev/911>911</a>: Felix Janda - Separate stat and statfs, give stat_main() a ds[2] array to distinguish FLAG_f vs not cases, and some whitespace changes.</li>
<li>commit <a href=/hg/toybox/rev/912>912</a>: description in commit message (also <a href=http://lists.landley.net/pipermail/toybox-landley.net/2013-May/001019.html>here</a>)</li>
<li><a href=http://lists.landley.net/pipermail/toybox-landley.net/2013-May/001024.html>design pondering</a> (leading to peek() function in lib/)</li>
<li>commit <a href=/hg/toybox/rev/914>914</a>: description in commit message.</li>
<li>commit <a href=/hg/toybox/rev/916>916</a>: description in commit message.</li>
<li>commit: <a href=/hg/toybox/rev/917>917</a>: description in commit message.</li>
<li>commit: <a href=/hg/toybox/rev/918>918</a>,
description: done: move pending to posix, default y, no other changes</a>.</li>
<li><a href=http://lists.landley.net/pipermail/toybox-landley.net/2013-June/001026.html>summary</a></li>
</ul>
<p>Status: COMPLETE.</p>
<!--#include file="footer.html" -->
|
