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