1 Open jobs for finishing GNU libc:
2 ---------------------------------
5 If you have time and talent to take over any of the jobs below please
6 contact <bug-glibc@gnu.org>.
8 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
10 [ 1] Port to new platforms or test current version on formerly supported
13 **** See http://www.gnu.org/software/libc/porting.html for more details.
16 [ 2] Test compliance with standards. If you have access to recent
17 standards (IEEE, ISO, ANSI, X/Open, ...) and/or test suites you
18 could do some checks as the goal is to be compliant with all
19 standards if they do not contradict each other.
22 [ 3] The IMHO opinion most important task is to write a more complete
23 test suite. We cannot get too many people working on this. It is
24 not difficult to write a test, find a definition of the function
25 which I normally can provide, if necessary, and start writing tests
26 to test for compliance. Beside this, take a look at the sources
27 and write tests which in total test as many paths of execution as
31 [ 4] Write translations for the GNU libc message for the so far
32 unsupported languages. GNU libc is fully internationalized and
33 users can immediately benefit from this.
35 Take a look at the matrix in
36 ftp://ftp.gnu.org/pub/gnu/ABOUT-NLS
37 for the current status (of course better use a mirror of ftp.gnu.org).
40 [ 6] Write `long double' versions of the math functions.
42 The libm is in fact fdlibm (not the same as in Linux libc 5).
44 **** Partly done. But we need someone with numerical experiences for
48 [ 7] Several math functions have to be written:
52 with long double arguments.
54 Beside this most of the complex math functions which are new in
55 ISO C99 should be improved. Writing some of them in assembler is
56 useful to exploit the parallelism which often is available.
59 [ 8] If you enjoy assembler programming (as I do --drepper :-) you might
60 be interested in writing optimized versions for some functions.
61 Especially the string handling functions can be optimized a lot.
65 Faster String Functions
66 Henry Spencer, University of Toronto
67 Usenix Winter '92, pp. 419--428
69 or just ask. Currently mostly i?86 and Alpha optimized versions
70 exist. Please ask before working on this to avoid duplicate
74 [10] Extend regex and/or rx to work with wide characters and complete
75 implementation of character class and collation class handling.
77 It is planned to do a complete rewrite.
79 *** We have now multibyte character support. But a rewrite is still
83 [11] Write access function for netmasks, bootparams, and automount
84 databases for nss_files and nss_db module.
85 The functions should be embedded in the nss scheme. This is not
86 hard and not all services must be supported at once.
89 [15] Cleaning up the header files. Ideally, each header style should
90 follow the "good examples". Each variable and function should have
91 a short description of the function and its parameters. The prototypes
92 should always contain variable names which can help to identify their
95 int foo (int, int, int, int);
99 *** The conformtest.pl tool helps cleaning the namespace. As far as
100 known the prototypes all contain parameter names. But maybe some
101 comments can be improved.
104 [16] The libio stream file functions should be extended in a way to use
105 mmap to map the file and use it as the buffer to user sees. For
106 read-only streams this should be rather easy and it avoids all read()
109 A more sophisticated solution would use mmap also for writing. The
110 standards do not demand that the file on the disk is always in the
111 correct form so it would be possible to enlarge it always according
112 to the page size and install the correct length only for fclose() and
116 [18] Based on the sprof program we need tools to analyze the output. The
117 result should be a link map which specifies in which order the .o
118 files are placed in the shared object. This should help to improve
119 code locality and result in a smaller foorprint (in code and data
120 memory) since less pages are only used in small parts.
123 [19] A user-level STREAMS implementation should be available if the
124 kernel does not provide the support.
126 *** This is a much lower priority job now that STREAMS are optional in
130 [20] More conversion modules for iconv(3). Existing modules should be
131 extended to do things like transliteration if this is wanted.
132 For often used conversion a direct conversion function should be
136 [21] The nscd program and the stubs in the libc should be changed so
137 that each program uses only one socket connect. Take a look at
138 http://www.cygnus.com/~drepper/nscd.html
140 An alternative approach is to use an mmap()ed file. The idea is
142 - the nscd creates the hash tables and the information it stores
143 in it in a mmap()ed region. This means no pointers must be
146 if POSIX shared memory is available use a named shared memory
147 region to put the data in
148 - each program using NSS functionality tries to open the file
150 - by checking some timestamp (which the nscd renews frequently)
151 the programs can test whether the file is still valid
152 - if the file is valid look through the nscd and locate the
153 appropriate hash table for the database and lookup the data.
154 If it is included we are set.
155 - if the data is not yet in the database we contact the nscd using
156 the currently implemented methods.
159 [22] It should be possible to have the information gconv-modules in
160 a simple cache which is faster to access. Using libdb is probably
161 overkill and loading it would probably be slower than reading the
162 plain text file. But a file format with a simple hash table and
163 some data it points to should be fine. Probably it should be
164 two tables, one for the aliases, one for the mappings. The code
165 should start similar to this:
167 if (stat ("gconv-modules", &stp) == 0
168 && stat ("gconv-modules.db", &std) == 0
169 && stp.st_mtime < std.st_mtime)
171 ... use the cache ...
175 ... use the plain file if it exists, otherwise the db ...
179 [23] The `strptime' function needs to be completed. This includes among
180 other things that it must get teached about timezones. The solution
181 envisioned is to extract the timezones from the ADO timezone
182 specifications. Special care must be given names which are used
183 multiple times. Here the precedence should (probably) be according
184 to the geograhical distance. E.g., the timezone EST should be
185 treated as the `Eastern Australia Time' instead of the US `Eastern
186 Standard Time' if the current TZ variable is set to, say,
187 Australia/Canberra or if the current locale is en_AU.
190 [25] Sun's nscd version implements a feature where the nscd keeps N entries
191 for each database current. I.e., if an entries lifespan is over and
192 it is one of the N entries to be kept the nscd updates the information
193 instead of removing the entry.
195 How to decide about which N entries to keep has to be examined.
196 Factors should be number of uses (of course), influenced by aging.
197 Just imagine a computer used by several people. The IDs of the current
198 user should be preferred even if the last user spent more time.
204 [27] We need a second test suite with tests which cannot run during a normal
205 `make check' run. This test suite can require root priviledges and
206 can test things like DNS (i.e., require network access),
207 user-interaction, networking in general, and probably many other things.