Debian Codegraph @

TODO

raw
1 finish up lexical warnings
2
3 omitting parked=>1 should be a warning, not an error?
4
5 demo/tail-f
6 see File::Tail or GNU textutils for implementation examples
7
8 add timeout for var watchers?
9
10 find owner for Win32 port
11
12 perhaps some help with distributing SIGCHLD?
13
14 documentation
15 how to design new event types (in perl)
16 how to design new event types (in C)??
17
18 [email protected] signal discussion:
19 The API rsignal overrides the current signal handler. If the signal
20 watcher is cancelled (or otherwise disabled) then rsignal is reset to
21 SIG_DFL. Maybe the original handler
22 should be restored. On the other hand, assignment to %SIG will mess
23 up Event watchers so maybe Event should refuse to watch a signal if a
24 handler is already installed.
25
26 optimization
27 update io event-mask without invalidating cache
28 determine IntervalEpsilon dynamically & make available from perl
29 use fancy profiler tricks to figure out how to boost performance
30 queue events from both sides?
31 try to hoist system calls from the inner loop
32
33 there seems to be some sort of bug in the shutdown phase
34 Attempt to free unreferenced scalar during global destruction.
35 Use of uninitialized value during global destruction.
36 Explicit blessing to '' (assuming package main) during global
37 destruction.
38
39 rethink
40 pe_check_recovery & ENTER/LEAVE points
41
42 increase the number of priority levels?
43 ASYNC 0 1 2 3 4 5 6 7 8 9 IDLE
44 HIGH NORM
45
46 loop
47 StarvePrio?
48
49 -----------------------------------------------------------------------
50 -----------------------------------------------------------------------
51
52 Move process.pm to external module?
53
54 accept "%f minutes" & "%f hours" for all time specifications?
55
56
57 On Wed, Aug 30, 2000 at 02:50:20PM -0400, Joshua N Pritikin wrote:
58 > Wow! I am reading about the Linux linked list implementation. In
59 > comparison to Linux, Event wastes an unnecessary pointer on every link.
60 > I'd better fix this! :-)
61
62 False alarm? Linux is able to make the assumption that the next & prev
63 pointers are always the same offset from the top of the structure. This
64 assumption is true for pe_watcher.all, pe_event.peer, pe_event.que, and
65 pe_qcallback.ring. However, pe_timeable.ring is stored at lots of
66 different offsets. Hm hm. I guess some bytes are wasted but otherwise
67 Event will need two different kinds of linked lists...