70322ae3 |
1 | TODO list for agedu |
2 | =================== |
3 | |
4 | Before it's non-embarrassingly releasable: |
5 | |
f2e52893 |
6 | - more flexible running modes |
7 | + combined scan+dump mode which doesn't even generate an index |
8 | file (nearly indistinguishable from find(1)) |
9 | + load mode which reads a dump from standard input and builds |
10 | the index (need to nail down a perfectly general dump format) |
11 | + at least some ability to chain actions within the same run: |
12 | "agedu -s dirname -w" would seem handy. |
70322ae3 |
13 | |
9d0b9596 |
14 | - work out what to do about atimes on directories in the absence of |
15 | the Linux syscall magic |
70322ae3 |
16 | * one option is to read them during the scan and reinstate them |
17 | after each recursion pop. Race-condition prone. |
9d0b9596 |
18 | * marking them in a distinctive colour in the reports is another |
19 | option. |
20 | * a third option is simply to ignore space taken up by |
21 | directories in the first place; inaccurate but terribly simple. |
22 | * incidentally, sometimes open(...,O_NOATIME) will fail, and |
23 | then we have to fall back to ordinary open. Be prepared to do |
24 | this, which probably means getting rid of the icky macro |
25 | hackery in du.c and turning it into a more sensible run-time |
26 | abstraction layer. |
70322ae3 |
27 | |
f2e52893 |
28 | - polish the plain-text output to make it look more like du |
29 | + configurable recursive output depth |
30 | + show the right bits last |
31 | |
32 | - figure out what to do about scans starting in the root directory |
33 | + Currently we end up with a double leading slash on the |
34 | pathnames, which is ugly, and we also get a zero-length href |
35 | in between those slashes which means the web interface doesn't |
36 | let you click back up to the top level at all. |
37 | + One big problem here is that a lot of the code assumes that |
38 | you can find the extent of a pathname by searching for "foo" |
39 | and "foo^A", trusting that anything inside the directory will |
40 | begin "foo/". So I'd need to consistently fix this everywhere |
41 | so that a trailing slash is disregarded while doing it, but |
42 | not actually removed. |
43 | + The text output gets it all wrong. |
44 | + The HTML output is fiddly even at the design stage: where |
45 | would I _ideally_ put the link to click on to get back to /? |
46 | It's unclear! |
47 | |
48 | - cross-Unix portability: |
49 | + use autoconf |
50 | * configure use of stat64 |
51 | * configure use of /proc/net/tcp |
52 | * configure use of /dev/random |
53 | * configure use of Linux syscall magic replacing readdir |
54 | + later glibcs have fdopendir, hooray! So we can use that |
55 | too, if it's available and O_NOATIME is too. |
56 | * what do we do elsewhere about _GNU_SOURCE? |
57 | |
58 | - prepare a little in advance for a potential future Windows port: |
373a02e5 |
59 | + store the path separator character in the index file when |
60 | writing it, and load it back in when reading |
f2e52893 |
61 | + store literal byte sizes in all the size fields, instead of |
62 | Unixoid 512-byte sectors |
70322ae3 |
63 | |
f2e52893 |
64 | - man page, licence. |
9d0b9596 |
65 | |
70322ae3 |
66 | Future directions: |
67 | |
1e8d78b9 |
68 | - IPv6 support in the HTTP server |
69 | |
70322ae3 |
70 | - run-time configuration in the HTTP server |
71 | * I think this probably works by having a configuration form, or |
72 | a link pointing to one, somewhere on the report page. If you |
73 | want to reconfigure anything, you fill in and submit the form; |
74 | the web server receives HTTP GET with parameters and a |
75 | referer, adjusts its internal configuration, and returns an |
76 | HTTP redirect back to the referring page - which it then |
77 | re-renders in accordance with the change. |
78 | * All the same options should have their starting states |
79 | configurable on the command line too. |
80 | |
70322ae3 |
81 | - curses-ish equivalent of the web output |
82 | + try using xterm 256-colour mode. Can (n)curses handle that? If |
83 | not, try doing it manually. |
f2e52893 |
84 | + I think my current best idea is to bypass ncurses and go |
85 | straight to terminfo: generate lines of attribute-interleaved |
86 | text and display them, so we only really need the sequences |
87 | "go here and display stuff", "scroll up", "scroll down". |
88 | + I think the attribute-interleaved text might be possible to do |
89 | cunningly, as well: we autodetect a basically VT-style |
90 | terminal, and add 256-colour sequences on the end. So, for |
91 | instance, we might set ANSI-yellow foreground, set ANSI-red |
92 | background, _then_ set both foreground and background to the |
93 | appropriate xterm 256-colour, and then display some |
94 | appropriate character which would have given the right blend |
95 | of the ANSI-16 fore and background colours. Then the same |
96 | display code should gracefully degrade in the face of a |
97 | terminal which doesn't support xterm-256. |
98 | * current best plan is to simulate the xterm-256 shading from |
99 | 0/5 to 5/5 by doing space, colon and hash in colour A on |
100 | colour B background, then hash, colon and space in B on A |
101 | background. |
102 | + Infrastructure work before doing any of this would be to split |
103 | html.c into two: one part to prepare an abstract data |
104 | structure describing an HTML-like report (in particular, all |
105 | the index lookups, percentage calculation, vector arithmetic |
106 | and line sorting), and another part to generate the literal |
107 | HTML. Then the former can be reused to produce very similar |
108 | reports in coloured plain text. |
70322ae3 |
109 | |
f2e52893 |
110 | - http://msdn.microsoft.com/en-us/library/ms724290.aspx suggest |
111 | modern Windowses support atime-equivalents, so a Windows port is |
373a02e5 |
112 | possible in principle. |
113 | + For a full Windows port, would need to modify the current |
114 | structure a lot, to abstract away (at least) memory-mapping of |
115 | files, details of disk scan procedure, networking for httpd. |
116 | Unclear what the right UI would be on Windows, too; |
117 | command-line exactly as now might be considered just a |
118 | _little_ unfriendly. Or perhaps not. |
119 | + Alternatively, a much easier approach would be to write a |
120 | Windows version of just the --scan-dump mode, which does a |
121 | filesystem scan via the Windows API and generates a valid |
122 | agedu dump file on standard output. Then one would simply feed |
123 | that over the network connection of one's choice to the rest |
124 | of agedu running on Unix as usual. |