4 - we should munmap in all operating modes where we mmapped,
5 otherwise chaining them will run out of address space
7 - we could still be using more of the information coming from
8 autoconf. Our config.h is defining a whole bunch of HAVE_FOOs for
9 particular functions (e.g. HAVE_INET_NTOA, HAVE_MEMCHR,
10 HAVE_FNMATCH). We could usefully supply alternatives for some of
11 these functions (e.g. cannibalise the PuTTY wildcard matcher for
12 use in the absence of fnmatch, switch to vanilla truncate() in
13 the absence of ftruncate); where we don't have alternative code,
14 it would perhaps be polite to throw an error at configure time
15 rather than allowing the subsequent build to fail.
16 + however, I don't see anything here that looks very
17 controversial; IIRC it's all in POSIX, for one thing. So more
18 likely this should simply wait until somebody complains.
20 - it would be useful to support a choice of indexing strategies.
21 The current system's tradeoff of taking O(N log N) space in order
22 to be able to support any age cutoff you like is not going to be
23 ideal for everybody. A second more conventional mechanism which
24 allows the user to specify a number of fixed cutoffs and just
25 indexes each directory on those alone would undoubtedly be a
26 useful thing for large-scale users. This will require
27 considerable thought about how to make the indexers pluggable at
28 both index-generation time and query time.
30 - IPv6 support in the HTTP server
31 * of course, Linux magic auth can still work in this context; we
32 merely have to be prepared to open one of /proc/net/tcp or
33 /proc/net/tcp6 as appropriate.
35 - run-time configuration in the HTTP server
36 * I think this probably works by having a configuration form, or
37 a link pointing to one, somewhere on the report page. If you
38 want to reconfigure anything, you fill in and submit the form;
39 the web server receives HTTP GET with parameters and a
40 referer, adjusts its internal configuration, and returns an
41 HTTP redirect back to the referring page - which it then
42 re-renders in accordance with the change.
43 * All the same options should have their starting states
44 configurable on the command line too.
46 - curses-ish equivalent of the web output
47 + try using xterm 256-colour mode. Can (n)curses handle that? If
48 not, try doing it manually.
49 + I think my current best idea is to bypass ncurses and go
50 straight to terminfo: generate lines of attribute-interleaved
51 text and display them, so we only really need the sequences
52 "go here and display stuff", "scroll up", "scroll down".
53 + Infrastructure work before doing any of this would be to split
54 html.c into two: one part to prepare an abstract data
55 structure describing an HTML-like report (in particular, all
56 the index lookups, percentage calculation, vector arithmetic
57 and line sorting), and another part to generate the literal
58 HTML. Then the former can be reused to produce very similar
59 reports in coloured plain text.
61 - http://msdn.microsoft.com/en-us/library/ms724290.aspx suggest
62 modern Windowses support atime-equivalents, so a Windows port is
63 possible in principle.
64 + For a full Windows port, would need to modify the current
65 structure a lot, to abstract away (at least) memory-mapping of
66 files, details of disk scan procedure, networking for httpd.
67 Unclear what the right UI would be on Windows, too;
68 command-line exactly as now might be considered just a
69 _little_ unfriendly. Or perhaps not.
70 + Alternatively, a much easier approach would be to write a
71 Windows version of just the --scan-dump mode, which does a
72 filesystem scan via the Windows API and generates a valid
73 agedu dump file on standard output. Then one would simply feed
74 that over the network connection of one's choice to the rest
75 of agedu running on Unix as usual.
~mdw / sgt / agedu / blob