summaryrefslogtreecommitdiffstats
path: root/TODO.dan
blob: a6c33682087b3d950e75e5a9e7e22eb05dc91285 (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
TODO.dan
========

This is my personal TODO list. No guarantees if it is out of date and things no
longer apply. If you want to help out with any of it, send me an email.

Pacman 3.1:
-----------

Downgrade feature - allow users to see cached packages and downgrade to
(previous or any?) available options.

Installed size and download size in -Si/Qi output should scale with package
size- KB to MB to GB. We should also get consistancy of K/KB, M/MB, etc.

Extreme similarity between some of the sync and add code...we have to be able
to abstract more away from sync actions and add actions to just 'actions'
(example: sync,c, add.c, and deptest.c all contain a switch on PM_DEP_MOD_*).
Merge update, add, and sync code when possible, so we reduce duplication.

Review progress/transaction event subsystem. (from TODO)

Add Doxygen comments to every function in libalpm including private functions.
Ensure functions are doing exactly their purpose and nothing more. Find
functions that perform duplicate behavior and merge them. Combine and refactor
others. Delete unnecessary functions. Stop keeping everything clustered in
little add/sync/remove parts, allow transactions to share code.

Go through options list. Decide if namings are good, all options are still
relevant, etc. Ideas for -Re (#6273), changing meaning of -c (has two meanings,
another FS bug), etc. Remove the -A flag and possibly -D, -T, and -Y (-Y is
killed now in favor of vercmp binary) if they can be done by other actions.
Possible switch of -U --> -I (#5571).

Review display and logging functions. There seems to be an abundance of them.
Make it extensible to use color, review what verbose/debug means. Perhaps
separate logging functionality- Pacman has its normal log, and alpm backend
keeps a very simple log file - listing only adds (including syncs) and removes.
This way a consistency list can be kept of what is currently installed and what
isn't, without all the logging junk from the front end.

Profile the code. Find the functions that are being called a lot, and see what
can be done about it. Find out if all these calls are necessary (e.g. excessive
alpm_list_count calls), and maybe think about changing data structures to speed
operations up (e.g. a 'count' field). NOTE: already found two huge issues with
quick profiling- excessive setlocale calls, and extremely excessive strcmp
calls.

Fix other localized issues- use non-printf when necessary. We may need to use
some wchar_t output on the progress bar as char/byte counts differ here. Sizes
of packages (e.g. 10,400.23 MB) should all be localized with correct
seperators.

Rewrite makepkg to use terminal-safe coloring/bolding. tput utility should
allow us to do this. Make universal message functions for systemwide use,
including all pacman utilites- abs, pacman-optimize, etc.

Bugs/FRs to smash: 6468, 6437, 6430?, 6420, 6404, 6389, 6312?, 6284, 6273?,
6255?, 6208, 5987, 5885, 5571, 4182, 3492, 2810?, 1769, 1588, 1571

Update doxygen comments, they may need some work. Try to document all of the
private internal functions too- it helps a ton for people just getting a
start on pacman hacking.

Other smaller things:
---------------------

unsigned int vs. unsigned- determine a standard and stick with it

FS #4185 - move where message is logged, perhaps?

Update copyrights (2007)

Update pacman website, and add/finish pacman coding style page

Refine makepkg error codes. Each kind of failure could have its own code:
  --package already built
  --failed integ checks
  --failed build
  --etc.

Add utility function to either frontend or backend to convert sizes: e.g. bytes
to KB, MB, GB.

Revamp the downloadprog function a bit. Seems kind of messy.

--print-uri option to sync should not require saying yes or no to up to date

Build a replacement for this, or at least standardize its use. We shouldn't
always need to pass handle->root around, it is constant. Something like char*
buildpath(file).
	/* build the new entryname relative to handle->root */
	snprintf(filename, PATH_MAX, "%s%s", handle->root, entryname);

Random ideas:
-------------

Possibly split utilities/extras from pacman package

Make .indent.pro file for GNU indent (started, but didn't finish)