summaryrefslogtreecommitdiff
path: root/Build/source/libs/zziplib/zziplib-0.13.59/TODO
blob: 1a408e4a12d4b28df0745ecfe7a01f7b19a21732 (plain)
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
SHORTTERM
- handle international filenames more gracefully (unicode API?)
- most is multithreaded ... but zzip_dir_open (Thorsten Schöning)
- rboerdijk@ does also report errors on overlapping reads, another
   one pointed to the usage of seek_set that may cause the problems

WISHLIST

- Check the CRC value at the end of read... and add more error codes.

- Do more test. Currently use only with tested "friendly" archives.
  This is also related to usages of zziplib in virus detection
  code which should better have a hardened library code. That does
  also include inflate interface code to need the most testing.

- the buffer reusage code was not strictly multithreading. It should
  be fixed by now but it would be better to have an automatic test
  routine to check reentrancy/multithreaded functionality.

- Sligthly More documentation. With the generation of man pages and
  multiple pages for the website, it does already look acceptable.
  It should still get better of course - kinda newbie friendly *g*

- Boris Schäling likes to open a zzip archive in memory.

KNOWN PROBLEMS

The win32 compilers need each a different config.h derivate that
matches both the headers shipped with the compiler and installed
with updates of the SDK. There is no autoconfigure on win32 as
that - unless you install some unix tools along.

The sparc-sun-solaris2.* will utter warnings for "char subscript"
which is caused by isdigit() from ctype.h - this will NOT FIX as
it is only in the example source code and we want to keep those
lean and mean to make them easy to adopt by developers.

The hppa1.1-hp-hpux10.20 did show spurious problems of making
shared libraries - this may well fix with an update of the
libtool package, the libtool 1.4 is dated 2001/04/24

There are reports of misaligned access to some zip fields that
I would guess to be on little-endian non-x86 platforms. The current
bytewise access of multibyte fields is targetted towards the
bigendian unix machines. The fix would need to go to fetch.h but
so far no response came about as that one could test a solution.

There are spurious reports of users on win32 platforms that tell
of some problems with a specific zip file they have but it was
not possible so far to recreate an environment abroad to show
the problem too. One can not say if that is due some general
instability out of DLL hell, or if there is a bug hiding somewhere.
Please send all those zip files to the maintainer, perhaps it
can help to find the real cause (I doubt it is in zziplib, but..)

Since lately the xml docbook tools have hardened the checks on the
input xml that is used for manpage generation. Interestingly the
resulting manpages are still okay but one should try to fixaway the
warnings as may be later the result would lead to garbage output
due more changes in the tools. Needs to change the xml generator
used in zzip (a python script).

TESTED PLATFORMS
sparc-sun-solaris2.6/gcc2.95.3
sparc-sun-solaris2.8/gcc2.95.3
hppa1.1-hp-hpux10.20
i686-mandrake-linux-9.0/gcc3.2
i686-mandrake-linux-9.1/gcc3.2.2
i686-debian-linux-2.2/gcc2.95.2
i386-unknown-freebsd4.7/gcc2.95.4          (formerly with wrapwrap)
powerpc-apple-darwin5.5                    (formerly with wrapwrap)
alphaev67-unknown-linux-gnu/gcc2.95.4      (that's a 64bit platform)
i386-ms-win32/msvc6
i386-ms-win32/msvc7
i386-ms-win32/mingw+msys
... and probably a lot of others not known to the maintainer.

Additionally, note that Sourceforge has discontinued their compilefarm
server laboratory. That makes it unlikely that proper support for
crossplatform functionality can be provided. Expect a compile problem
here or there - the code however should be prepared to get around any
problems easily. Send patches! (especially Linux distributions makers
are usually not sending their patches to upstream maintainers).

Note: the latest cross platform tests are done indirectly by using
the build.opensuse.org rpm packaging where one can run "make check"
just before doing the "make install" of the compiled library.


SUSE BUILDSERVER INFO

I: A function overflows or underflows an array access. This could be a real error,
but occasionaly this condition is also misdetected due to loop unrolling or strange pointer
handling. So this is warning only, please review.
W: zziplib arraysubscript ../../zzip/memdisk.c:114

I: File is compiled without RPM_OPT_FLAGS
W: zziplib no-rpm-opt-flags <cmdline>:../../SDL/SDL_rwops_zzcat.c, ../../SDL/SDL_rwops_zzip.c

I: Program is likely to break with new gcc. Try -fno-strict-aliasing.
W: zziplib strict-aliasing-punning ../../zzip/file.c:275
W: zziplib strict-aliasing-punning ../../zzip/fseeko.c:99, 126, 147, 158, 182, 281, 288, 301, 360, 539, 543, 546, 563
W: zziplib strict-aliasing-punning ../../zzip/memdisk.c:181, 182, 183, 185, 186, 187, 188, 189, 192, 193, 194, 195, 247, 455, 456
W: zziplib strict-aliasing-punning ../../zzip/mmapped.c:277, 289, 311, 314, 339, 340, 393, 397, 410, 438, 440, 443, 444, 549, 551, 552, 558, 559, 561
W: zziplib strict-aliasing-punning ../../zzip/zip.c:318, 320, 321, 322, 339, 341, 342, 343, 484, 485, 486, 497, 498, 499, 500, 501
E: zziplib 64bit-portability-issue ../../zzip/memdisk.c:112
System halted.