1 * make nomtime work from parent dir
3 * test for buffer overflows when reading from global
4 buffer into special ones (item, etc).
5 * test whether nomtime can handle complicated filenames
6 * split into subdirectories
7 * split acceptance test into seperate dir,
8 tests into seperate files
9 * test driven refactoring
11 * replace assertions by proper error handling
12 * feed name is cutted if too long (replace using full stops)
13 * feed/title/desc injection not possible
14 * check whether new items are retrieved, but old ones are
16 * check whether dates are handled properly (timezones!)
17 * fix remaining bugs (see README)
19 * item overwrite only if date >= olddate
20 else notice, skip offending item, and do not change
21 feed date, continue but exit with error
22 (invariant: date of every unfetched item is > feed date)
23 * Make directories/items read-only.
24 * support lastBuildDate and pubDate for channels
25 * do not stop on first old item, consume whole feed
26 channel pubDate is not old.
27 * maybe use http last-modified? it is not reliable on
28 some servers, though, and probably needs ugly hacks
30 * we're currently not just aggregating, we're aggregating
32 * Drop HTML formatting!
35 |-- news item title, probably truncated or hashed?
37 | |-- description (opt)
41 * now *every* interface can be built on top.