1 TODO list for every uClibc release:
2 -------------------------------------------------
3 *) Test cris, i386, mips, mipsel, sh, x86_64, arm, armeb, and powerpc
4 with the latest LTP testsuite. Fix any regressions and post LTP
5 testsuite results for each architecture on uclibc.org.
6 *) Audit header files. Remove prototypes for all functions that
7 are not supported -- especially needed for the libm headers.
8 *) Audit header files. When options are disabled, also disable
9 them in the include files as well by checking for the proper
10 define from include/bits/uClibc_config.h (pulled in from features.h)
12 TODO list for the uClibc 0.9.31 release:
13 -------------------------------------------------
16 support arches: (- todo; + done)
24 TOC: http://www.opengroup.org/onlinepubs/9699919799/xrat/contents.html
25 shell (busybox): http://www.opengroup.org/onlinepubs/9699919799/xrat/V4_xcu_chap01.html#tag_22_01_01
27 http://www.opengroup.org/onlinepubs/9699919799/xrat/V4_xbd_chap13.html#tag_21_13_02
28 http://www.opengroup.org/onlinepubs/9699919799/xrat/V4_xsh_chap01.html#tag_23_01_01
31 TODO list for the uClibc 0.9.29 release:
32 -------------------------------------------------
33 *) as many of the arch-specific issues as possible
34 *) Remove N instances of libc_hidden_proto() from uClibc internals.
35 Instead add internal only header(s) defining all hidden prototypes.
36 This will avoid clutter and guarantee prototype consistancy.
37 *) The __is*_l() functions were all removed, such that we now only export
38 the is*_l() functions (no prefix). Before, we had the prefixed versions
39 for use by libstdc++ and weak versions without prefixes exported because
40 those functions belong to no std (unless you call glibc a std). This should
41 be fixed. Similar problems likely were created elsewhere.
43 http://bugs.uclibc.org/view.php?id=420
44 http://bugs.uclibc.org/view.php?id=539
45 *) bug in getopt handling:
46 http://bugs.uclibc.org/view.php?id=61
47 http://www.uclibc.org/lists/uclibc/2006-January/013851.html
48 *) Should integrate test subdir better ... need to propagate CPU
49 CFLAGS/LDFLAGS to the build/link for target binaries so that when we have
50 a multilib toolchain, the proper ABI is selected.
53 TODO list for the uClibc 1.0.0 release:
54 -------------------------------------------------
55 *) glob / fnmatch tests fail
56 *) regex should pass AT&T conformance tests
57 *) Finish hiding uClibc internal symbols from our exported namespace
58 *) Add system for generating minimal system call asm wrappers that abuse
59 the C ABI to minimize amount of register/stack manipulation
60 *) Explicity add some sortof CONFIG_LINUX_2_2_ABI, CONFIG_LINUX_2_4_ABI
61 and CONFIG_LINUX_2_6_ABI type options, rather than having the abi
62 depend on the selected set of kernel headers. This will likely also
63 require that we provide a set of kernel headers (probably a derivitive
64 of Mazur's linux-libc-headers) that can supplies the 2.2, 2.4, and 2.6
66 *) Documentation updates:
67 *) Write a uClibc HOWTO document
68 *) Update README document
69 *) Update INSTALL document
70 *) Update docs/Glibc_vs_uClibc_Differences.txt document
71 and fully document all differences between the feature
72 set of uClibc and glibc.
73 *) Update docs/uClibc_vs_SuSv3.txt document
74 *) Update docs/threads.txt document
75 *) Write man pages for ldd and ldconfig utility binaries
76 *) Implement some mechanism (perhaps encoded in the .so name,
77 perhaps using an abi tag magically embedded into each object)
78 for flagging config options that break the ABI. Options
79 such as UCLIBC_HAS_SOFT_FLOAT, UCLIBC_HAS_THREADS, UCLIBC_HAS_LOCALE,
80 and perhaps others (finalize list) produce a lib with a differing
81 ABI. Make it so apps cannot use an ABI mis-matched uClibc.
82 This is most easily done using symbol versioning...
83 *) Implement the long double versions of math funcs
84 using wrappers on top of the double versions (size / precision
85 trade off where size clearly wins).
86 *) Make all small objects (>~50 bytes) into either inlines or
88 *) Cleanup / rewrite sysconf.c. It should get some information
89 from ldso (such as HZ). Other stuff it currently just makes
90 up, which is obviously wrong. Also bits/uClibc_clk_tck.h
91 needs to be updated at the same time to get proper HZ values.
92 *) poll emulation using select() for old 2.0.x uClinux kernels
93 in libc/sysdeps/linux/common/poll.c fails some python self-tests.
94 Of course, modern systems using the actuall poll() syscall work fine.
95 *) Cleanup/scrub all the Makefile copyright junk
96 *) Fix dlopen, for both static and dynamic cases, and make it
97 fully comply with SuSv3
100 TODO list for AFTER the uClibc 1.0.0 release:
101 -------------------------------------------------
102 *) Add support for 64bit time (to deal withhttp://2038bug.com/):
103 date -s 011903142038 sets epoch to roll over in a few seconds (2^31)
104 *) Add support for Linux 2.6.x fast vsyscalls
105 *) Enable pristine source tree builds
106 *) Fix regex so it isn't so stinking big
107 *) Fix glob so it isn't so stinking big
108 *) run 'nm -D --size-sort -t d libuClibc-0.9.26.so' and work on the
109 biggest things (i.e. stuff at the end of the list) to make
111 <more wishlist items here>
115 -------------------------------------------------
117 ldso is not implemented
119 update crt1 code again for nommu (http://bugs.uclibc.org/view.php?id=538)
120 recruit jbowler and integrate thumb support
122 needs to be updated for the new ldso changes and unification of the
123 uClibc_main funcs (__uClibc_start_main doesnt exist anymore)
125 errno.c test segfaults causes segfault in clone.S (seems to be in glibc too)
127 add support for fast system calls
129 ldso is not implemented
131 ldso lazy relocation doesnt work
133 ldso is reasonably broken, presently requiring additional
136 need a sigaction.c since common one doesnt work (signal tests)
137 ldso needs to be updated since it's totally broken atm
139 pthread support, linktime warning support (implies GAS patches),
140 general touchups, testing, ldso.
143 -----------------------------------------------------------------------------
146 1) Little things that need fixing:
147 ----------------------------------
148 a) Fix bug in *printf: outdigit precison bug
149 b) Check that gnu/bsd extension members tm_gmtoff and tm_zone in struct tm
150 are respected where they should be.
151 c) Implement the obstack printf funcs for glibc compat.
152 d) Implement glibc 'a' flag for scanf string conversions.
153 e) Allow use of the older non-table-based ctype functions when using
154 stub locale support. (smaller)
155 f) __drand48_iterate should be void
156 g) alphasort vs. versionsort. The former seems to be SVID, the latter GNU
157 i.e. reverse to what we currently do. The latter is unimplemented.
158 h) ponder removal/configs to turn off: __xpg_*, bsd_signal, dysize,
161 2) Additional str{f|p}time issues.
162 ----------------------------------
163 a) Spacing issue wrt strptime.
164 b) Support locale specific alternate digits. (data is in place)
165 c) Support locale era in year designations. (data is in place)
166 d) Deal with mb format string issues in strftime.
167 e) Implement wcsftime.
169 3) Other locale issues (my implementation):
170 -------------------------------------------
171 a) Do a little more clean up of ctype and wctype.
172 b) Rework of the locale data organization to make using locales reasonable
173 when staticly linking. (mmap)
174 c) Rewrite the locale data generation tools to process the text specifications
175 rather than relying on glibc.
176 d) Adapt regex lib to use my collation data and add the necessary collating
177 item tables to support SUSv3 required features.
178 e) transliteration of unsupported wchars in 8-bit locales (like glibc).
179 f) Support ISO/IEC 14652 draft locale extensions (LC_PAPER, etc).
180 g) Implement strfrom.
181 h) Shift-state codeset locale support?
185 a) Port uClibc to other OSs (including elks), or even bare metal (libgloss).
186 b) Write a space-efficient gettext substitute, to avoid storing large amounts