3 Stuff that needs to be done. All of this is fair game for 1.2.
6 make -j is broken, -j1 is forced atm
9 doesn't understand (), lots of susv3 stuff.
12 The command shell situation is a big mess. We have three or four different
13 shells that don't really share any code, and the "standalone shell" doesn't
14 work all that well (especially not in a chroot environment), due to apps not
15 being reentrant. Unifying the various shells and figuring out a configurable
16 way of adding the minimal set of bash features a given script uses is a big
17 job, but it would be a big improvement.
19 Note: Rob Landley (rob@landley.net) is working on a new unified shell called
20 bbsh, but it's a low priority...
23 We should have a diff -u command. We have patch, we should have diff
24 (we only need to support unified diffs though).
26 Also, make sure we handle empty files properly:
27 From the patch man page:
29 you can remove a file by sending out a context diff that compares
30 the file to be deleted with an empty file dated the Epoch. The
31 file will be removed unless patch is conforming to POSIX and the
32 -E or --remove-empty-files option is not given.
35 Should have simple fuzz factor support to apply patches at an offset which
36 shouldn't take up too much space.
38 And while we're at it, a new patch filename quoting format is apparently
39 coming soon: http://marc.theaimsgroup.com/?l=git&m=112927316408690&w=2
42 It would be nice to have a man command. Not one that handles troff or
43 anything, just one that can handle preformatted ascii man pages, possibly
44 compressed. This could probably be a script in the extras directory that
45 calls cat/zcat/bzcat | less
47 (How doclifter might work into this is anybody's guess.)
50 Compression-side support.
63 bb_close() with fsync()
64 We should have a bb_close() in place of normal close, with a CONFIG_ option
65 to not just check the return value of close() for an error, but fsync().
66 Close can't reliably report anything useful because if write() accepted the
67 data then it either went out to the network or it's in cache or a pipe
68 buffer. Either way, there's no guarantee it'll make it to its final
69 destination before close() gets called, so there's no guarantee that any
70 error will be reported.
72 You need to call fsync() if you care about errors that occur after write(),
73 but that can have a big performance impact. So make it a config option.
76 Look at the full Single Unix Specification version 3 (available online at
77 "http://www.opengroup.org/onlinepubs/009695399/nfindex.html") and
78 figure out which of our apps are compliant, and what we're missing that
79 we might actually care about.
81 Even better would be some kind of automated compliance test harness that
82 exercises each command line option and the various corner cases.
85 How much internationalization should we do?
87 The low hanging fruit is UTF-8 character set support. We should do this.
88 (Vodz pointed out the shell's cmdedit as needing work here. What else?)
90 We also have lots of hardwired english text messages. Consolidating this
91 into some kind of message table not only makes translation easier, but
92 also allows us to consolidate redundant (or close) strings.
94 We probably don't want to be bloated with locale support. (Not unless we can
95 cleanly export it from our underlying C library without having to concern
96 ourselves with it directly. Perhaps a few specific things like a config
97 option for "date" are low hanging fruit here?)
99 What level should things happen at? How much do we care about
100 internationalizing the text console when X11 and xterms are so much better
101 at it? (There's some infrastructure here we don't implement: The
102 "unicode_start" and "unicode_stop" shell scripts need "vt-is-UTF8" and a
103 --unicode option to loadkeys. That implies a real loadkeys/dumpkeys
104 implementation to replace loadkmap/dumpkmap. Plus messing with console font
105 loading. Is it worth it, or do we just say "use X"?)
108 Lots of archivers have the same general infrastructure. The directory
109 traversal code should be factored out, and the guts of each archiver could
110 be some setup code and a series of callbacks for "add this file",
111 "add this directory", "add this symlink" and so on.
113 This could clean up tar and zip, and make it cheaper to add cpio and ar
114 write support, and possibly even cheaply add things like mkisofs or
115 mksquashfs someday, if they become relevant.
118 Several existing applets (sort, vi, less...) read
119 a whole file into memory and act on it. There might be an opportunity
120 for shared code in there that could be moved into libbb...
122 Individual compilation of applets.
123 It would be nice if busybox had the option to compile to individual applets,
124 for people who want an alternate implementation less bloated than the gnu
125 utils (or simply with less political baggage), but without it being one big
128 Turning libbb into a real dll is another possibility, especially if libbb
129 could export some of the other library interfaces we've already more or less
130 got the code for (like zlib).
132 buildroot - Make a "dogfood" option
133 Busybox 1.1 will be capable of replacing most gnu packages for real world use,
134 such as developing software or in a live CD. It needs wider testing.
136 Busybox should now be able to replace bzip2, coreutils, e2fsprogs, file,
137 findutils, gawk, grep, inetutils, less, modutils, net-tools, patch, procps,
138 sed, shadow, sysklogd, sysvinit, tar, util-linux, and vim. The resulting
139 system should be self-hosting (I.E. able to rebuild itself from source code).
140 This means it would need (at least) binutils, gcc, and make, or equivalents.
142 It would be a good "eating our own dogfood" test if buildroot had the option
143 of using a "make allyesconfig" busybox instead of the all of the above
144 packages. Anything that's wrong with the resulting system, we can fix. (It
145 would be nice to be able to upgrade busybox to be able to replace bash and
146 diffutils as well, but we're not there yet.)
148 One example of an existing system that does this already is Firmware Linux:
149 http://www.landley.net/code/firmware
152 Busybox should have a sample initramfs build script. This depends on
153 involves bbsh, mdev, and switch_root.
156 We have a CONFIG_BUFFER mechanism that lets us select whether to do memory
157 allocation on the stack or the heap. Unfortunately, we're not using it much.
158 We need to audit our memory allocations and turn a lot of malloc/free calls
159 into RESERVE_CONFIG_BUFFER/RELEASE_CONFIG_BUFFER.
161 And while we're at it, many of the CONFIG_FEATURE_CLEAN_UP #ifdefs will be
162 optimized out by the compiler in the stack allocation case (since there's no
163 free for an alloca()), and this means that various cleanup loops that just
164 call free might also be optimized out by the compiler if written right, so
165 we can yank those #ifdefs too, and generally clean up the code.
167 Switch CONFIG_SYMBOLS to ENABLE_SYMBOLS
169 In busybox 1.0 and earlier, configuration was done by CONFIG_SYMBOLS
170 that were either defined or undefined to indicate whether the symbol was
171 selected in the .config file. They were used with #ifdefs, ala:
179 In 1.1, we have new ENABLE_SYMBOLS which are always defined (as 0 or 1),
180 meaning you can still use them for preprocessor tests by replacing
181 "#ifdef CONFIG_SYMBOL" with "#if ENABLE_SYMBOL". But more importantly, we
182 can use them as a true or false test in normal C code:
184 if (ENABLE_SYMBOL && other_test) {
188 (Optimizing away if() statements that resolve to a constant value
189 is known as "dead code elimination", an optimization so old and simple that
190 Turbo Pascal for DOS did it twenty years ago. Even modern mini-compilers
191 like the Tiny C Compiler (tcc) and the Small Device C Compiler (SDCC)
192 perform dead code elimination.)
194 Right now, busybox.h is #including both "config.h" (defining the
195 CONFIG_SYMBOLS) and "bb_config.h" (defining the ENABLE_SYMBOLS). At some
196 point in the future, it would be nice to wean ourselves off of the
197 CONFIG versions. (Among other things, some defective build environments
198 leak the Linux kernel's CONFIG_SYMBOLS into the system's standard #include
199 files. We've experienced collisions before.)
202 This is more an unresolved issue than a to-do item. More thought is needed.
204 Normally we rely on exit() to free memory, close files, and unmap segments
205 for us. This makes most calls to free(), close(), and unmap() optional in
206 busybox applets that don't intend to run for very long, and optional stuff
207 can be omitted to save size.
209 The idea was raised that we could simulate fork/exit with setjmp/longjmp
210 for _really_ brainless embedded systems, or speed up the standalone shell
211 by not forking. Doing so would require a reliable FEATURE_CLEAN_UP.
212 Unfortunately, this isn't as easy as it sounds.
214 The problem is, lots of things exit(), sometimes unexpectedly (xmalloc())
215 and sometimes reliably (bb_perror_msg_and_die() or show_usage()). This
216 jumps out of the normal flow control and bypasses any cleanup code we
217 put at the end of our applets.
219 It's possible to add hooks to libbb functions like xmalloc() and bb_xopen()
220 to add their entries to a linked list, which could be traversed and
221 freed/closed automatically. (This would need to be able to free just the
222 entries after a checkpoint to be usable for a forkless standalone shell.
223 You don't want to free the shell's own resources.)
225 Right now, FEATURE_CLEAN_UP is more or less a debugging aid, to make things
226 like valgrind happy. It's also documentation of _what_ we're trusting
227 exit() to clean up for us. But new infrastructure to auto-free stuff would
228 render the existing FEATURE_CLEAN_UP code redundant.
230 For right now, exit() handles it just fine.
235 watchdog.c could autodetect the timer duration via:
236 if(!ioctl (fd, WDIOC_GETTIMEOUT, &tmo)) timer_duration = 1 + (tmo / 2);
237 Unfortunately, that needs linux/watchdog.h and that contains unfiltered
238 kernel types on some distros, which breaks the build.
243 Replace deprecated functions.
247 sigblock(), siggetmask(), sigsetmask(), sigmask() -> sigprocmask et al