1 |
20 |
jlechner |
std::allocator
|
2 |
|
|
|
3 |
|
|
- switch to mt_allocator with --enable-threads=posix.
|
4 |
|
|
|
5 |
|
|
- persistent allocator
|
6 |
|
|
|
7 |
|
|
- global/extern allocator
|
8 |
|
|
|
9 |
|
|
std::string
|
10 |
|
|
|
11 |
|
|
- Policy-based design incorporating COW
|
12 |
|
|
vs. deep copy issues, MT scalability
|
13 |
|
|
See Andrei Alexandrescu, June 2001, C/C++ Users Journal
|
14 |
|
|
"Generic: A Policy-Based basic_string Implementation"
|
15 |
|
|
http://www.cuj.com/documents/s=7994/cujcexp1906alexandr/
|
16 |
|
|
|
17 |
|
|
- operator!= and utility/rel_ops operators need to be made safe with
|
18 |
|
|
string and vector iterator classes. basic_string::reverse_iterator may
|
19 |
|
|
be implemented incorrectly, or need things like
|
20 |
|
|
operator==(__normal_iterator, const char*&), and swap(vector)
|
21 |
|
|
|
22 |
|
|
- 'do the right thing' ctor fixing needs to be done for string. This
|
23 |
|
|
is still subject to some debate on the library issues list, so I
|
24 |
|
|
suggest punting till the dust clears.
|
25 |
|
|
|
26 |
|
|
- fix template members of basic_string<> to overload iterators and
|
27 |
|
|
non-iterators properly. (This is the infamous hack as in vector<> etc
|
28 |
|
|
23.1.1 para 10.)
|
29 |
|
|
|
30 |
|
|
std::locale
|
31 |
|
|
|
32 |
|
|
- implement __convert_to_v and __convert_from_v without "C" library
|
33 |
|
|
functions and and LANG environment variable dependencies.
|
34 |
|
|
|
35 |
|
|
- use localedata to implement generic named (non-MT-safe) locales?
|
36 |
|
|
Figure out a way to use ICU data, like libjava? Re-package and use
|
37 |
|
|
the glibc localedata, even if we aren't on linux? Need a generic
|
38 |
|
|
locale model that does something besides the "C" locale.
|
39 |
|
|
|
40 |
|
|
- make locale::classic() separate from named locale code. This will
|
41 |
|
|
improve the static linkage situation, but will require new
|
42 |
|
|
initialization code. In particular, we need lazy-initialization of
|
43 |
|
|
locale::classic(), and maybe the has_facet/use_facet functions for all
|
44 |
|
|
the required facets. The end goal is a self-contained
|
45 |
|
|
locale_init.cc, or one with transitive closure without the locale
|
46 |
|
|
instantiations (locale-inst.cc) or the named locale bits
|
47 |
|
|
(localename.cc).
|
48 |
|
|
|
49 |
|
|
- Jerry(?)/Paolo(?) work on __float_to_char.
|
50 |
|
|
|
51 |
|
|
- minimize ctype convertion in data facets, see numpunct/num_put/num_get
|
52 |
|
|
|
53 |
|
|
std::basic_filebuf, 27_io
|
54 |
|
|
|
55 |
|
|
- wfilebuf, get variable-encoding working and tested, including
|
56 |
|
|
positioning and seeking. (I think this may be done now)
|
57 |
|
|
|
58 |
|
|
- wfilebuf testsuite (getting there...)
|
59 |
|
|
|
60 |
|
|
- look ahead for unbuffered io, so know when multiple putc's can be
|
61 |
|
|
coalesced.
|
62 |
|
|
|
63 |
|
|
- unlocked __basic_file + new mutext class
|
64 |
|
|
|
65 |
|
|
- optimized the sentries for istream/ostream
|
66 |
|
|
|
67 |
|
|
- v2 vs. v3 speed
|
68 |
|
|
|
69 |
|
|
- add optimization hooks (esp. whitespace eating) to streambuf
|
70 |
|
|
- add _M_begin() and _M_end() to streambuf
|
71 |
|
|
- add algorithm specializations for [io]streambuf_iterator (copy find etc.)
|
72 |
|
|
|
73 |
|
|
testsuite
|
74 |
|
|
|
75 |
|
|
- valgrind hooks into make check so can tell memory leakage
|
76 |
|
|
Some commentary on the valgrind users list
|
77 |
|
|
|
78 |
|
|
- add hooks for qmtest for visual diffs
|
79 |
|
|
|
80 |
|
|
- automatic testing of interactive tests
|
81 |
|
|
|
82 |
|
|
- diffing generated output files
|
83 |
|
|
|
84 |
|
|
- provide testsuites for numerics.
|
85 |
|
|
|
86 |
|
|
- make check-abi needs to have full symbol checking. Scope the LSB
|
87 |
|
|
testsuite, see what's going on with the typeinfo etc. bits.
|
88 |
|
|
|
89 |
|
|
- try to do a better job of ABI testing, with instantiations of all
|
90 |
|
|
standard-specified types checked, not just exported symbols.
|
91 |
|
|
|
92 |
|
|
g++/binutils
|
93 |
|
|
|
94 |
|
|
- compression for wide versions of basic types, not just narrow
|
95 |
|
|
|
96 |
|
|
threads
|
97 |
|
|
|
98 |
|
|
- create MT abstraction layer for atomicity to pthreads.
|
99 |
|
|
|
100 |
|
|
- solution for threads + C++.
|
101 |
|
|
|
102 |
|
|
- audit for places where __builtin_expect can be used.
|
103 |
|
|
|
104 |
|
|
- relocations, work on getting these down
|
105 |
|
|
|
106 |
|
|
- issues with __builtin_memcpy and std::copy from Jerry Quinn
|
107 |
|
|
http://gcc.gnu.org/ml/libstdc++/2003-02/msg00056.html
|
108 |
|
|
http://gcc.gnu.org/ml/libstdc++/2003-02/msg00302.html
|
109 |
|
|
http://gcc.gnu.org/ml/gcc/2003-10/msg01305.html
|
110 |
|
|
|
111 |
|
|
- fix dependency tracking for includes (.h, .tcc) during build process.
|
112 |
|
|
|
113 |
|
|
- coordinate with "C" library people the "C" compatibility headers.
|
114 |
|
|
|
115 |
|
|
- Think about naming all member data and member functions consistently
|
116 |
|
|
as per
|
117 |
|
|
funtions: _M_verb_adverb
|
118 |
|
|
data: _M_noun_adjective
|
119 |
|
|
|
120 |
|
|
- A C++STYLE guide that deals with nested namespaces, and that
|
121 |
|
|
everybody can live with.
|
122 |
|
|
|
123 |
|
|
- exception specifications need to be reviewed for all parts of the
|
124 |
|
|
library support and utility areas, particularly . Part of this is
|
125 |
|
|
a standards issue, where the 27_io bugs/standard is really in an odd
|
126 |
|
|
spot.
|
127 |
|
|
|
128 |
|
|
- C-related issues WRT to io and filepos, mbstate_t. Seeking in wide
|
129 |
|
|
streams. May need to define operators for mbstate_t so that
|
130 |
|
|
'mbstate_t& == mbstate_t' is something that can be done.
|
131 |
|
|
|
132 |
|
|
- scoping/linking issues WRT to C structs need to be worked out. See
|
133 |
|
|
Nathan's commentary on cantrip, http://www.cantrip.org/cheaders.html
|
134 |
|
|
|
135 |
|
|
- auto_ptr: seems to be some disagreement on what is
|
136 |
|
|
standards-conformant behavior, specially on conversion operators.
|
137 |
|
|
|
138 |
|
|
- list::assignment operator needs const_cast
|
139 |
|
|
|
140 |
|
|
- a cleaner division between pointers-to-value_type and true iterators
|
141 |
|
|
needs to be drawn throughout the entire STL implementation.
|
142 |
|
|
|
143 |
|
|
- priority_queue conversions may be non-conformant
|
144 |
|
|
|
145 |
|
|
- Protect valarray::result_type (not Standard) and make it work with
|
146 |
|
|
the various helper classes.
|
147 |
|
|
|
148 |
|
|
- Make sure `valarray & == _Expr<_BinClos,bool>'
|
149 |
|
|
is defined
|
150 |
|
|
|
151 |
|
|
- All of the Library working group closed issues need to be
|
152 |
|
|
addressed. Some of them proposed resolutions are already in the v-3
|
153 |
|
|
sources, with macro-guards. Also, same with the TR.
|
154 |
|
|
|
155 |
|
|
- add deprecated features from Annex D
|
156 |
|
|
- add #ifdef wrappers for all of them, including in each
|
157 |
|
|
C header
|
158 |
|
|
- replace typedef'd non-deprecated types with more type-safe equivalents.
|
159 |
|
|
|
160 |
|
|
- add feature-test macros for non-standard extensions
|
161 |
|
|
|
162 |
|
|
- add MT support for locale, string, istream, ostream
|
163 |
|
|
|
164 |
|
|
- need to think about doing a .texi or DocBook manual, instead of all
|
165 |
|
|
these HTML pages. In addition, it would be nice to have a full manual,
|
166 |
|
|
instead of a lot of ad-hoc pages. Weaknesses include numerics, locale,
|
167 |
|
|
and io.
|
168 |
|
|
|
169 |
|
|
- add FAQ entries -- improve the install instructions
|
170 |
|
|
|
171 |
|
|
- add HOWTO entries
|
172 |
|
|
|
173 |
|
|
- do more doxygen manpages
|
174 |
|
|
|