| 1 | Revision history for AnyEvent::MP |
1 | Revision history for AnyEvent::MP |
| 2 | |
2 | |
| 3 | TODO: testsuite |
3 | TODO: testsuite |
| 4 | TODO: intro: maybe simple job pool example? |
4 | TODO: intro: maybe simple job pool example? |
| 5 | TODO: intro: mention watchdog... |
5 | TODO: intro: mention watchdog... |
| 6 | TODO: use gvpe method of making contact, if possible. |
6 | TODO: maybe disbale warnings by default? |
|
|
7 | TODO: listener-scopes (10.0.0.1:4040@vpn) and connect-scopes ("vpn,public") |
|
|
8 | |
|
|
9 | faq: can't se anything |
|
|
10 | faq: all is asynch |
|
|
11 | faq: how to interface to non-perl nodes? |
|
|
12 | |
|
|
13 | TODO: check gproto, nproto, on connect |
|
|
14 | TODO: eg/* |
|
|
15 | TODO: #TODO##d# markers |
|
|
16 | TODO: limiting reconnecting speed when unreachable? |
|
|
17 | 2 |
|
|
18 | - INCOMPATIBLE CHANGE: the node protocol used by this version is not |
|
|
19 | compatible to versions <2 - you have to upgrade all nodes. |
|
|
20 | - INCOMPATIBLE CHANGE: many APIs have changed, see the section |
|
|
21 | "PORTING FROM VERSION 1" in the AnyEvent::MP manpage for details. |
|
|
22 | - AnyEvent::MP::Global is now optional - only seed nodes autoload it |
|
|
23 | and create a fully-meshed network with each other, normal nodes only |
|
|
24 | have to connect to seed nodes, other connections are on-demand. |
|
|
25 | - new "norc" and "force" options for MP::configure to ignore |
|
|
26 | or override the rc file. |
|
|
27 | - move seed code to Kernel. |
|
|
28 | - non-json receive framing was probably broken. |
|
|
29 | - fixed small typo in MP::Transport. |
|
|
30 | - fix (unused) provider-kv (version sometimes missing). |
|
|
31 | - reduce the default export list of MP::Kernel. |
|
|
32 | - reduced various random delays to zero, as they |
|
|
33 | are hopefully no longer needed with the new global |
|
|
34 | protocol. |
|
|
35 | - implement a secure mode that can suppress code execution |
|
|
36 | requests on a per-node/per-request basis. |
|
|
37 | - do not use GMP for generating alphanumeric strings even when |
|
|
38 | available, use a more tricky algorithm instead. |
|
|
39 | - anonymous node names and remote port names have been shortened |
|
|
40 | considerably. |
|
|
41 | - $AnyEvent::MP::Kernel::SRCNODE contains the node ID of the origin |
|
|
42 | node during message receives. |
|
|
43 | - aemp shell now supports "package" selection and offers an |
|
|
44 | $ECHO port you can send messages to. |
|
|
45 | - rely on perl rand() instead of using /dev/urandom when available, as |
|
|
46 | hopefully it is random enough. |
|
|
47 | - aemp shell no longer leaks one port per command. |
|
|
48 | - nodenames can contain %-escapes such as %n or %u. |
|
|
49 | - aemp now uses aemp/%n/%u as nodename. |
|
|
50 | - configure supports "eval", a small perl snippet to initialsie a node, |
|
|
51 | most useful in "aemp run eval ...". |
|
|
52 | - known_nodes is gone, it has little value, use all_nodes as |
|
|
53 | replacement. |
|
|
54 | - lots of small change and improvements. |
|
|
55 | - node_is_known has been removed. |
|
|
56 | - TRACE now shows the portid on SND. |
|
|
57 | |
|
|
58 | 1.30 Thu Jun 30 11:30:39 CEST 2011 |
|
|
59 | - connection errors at the right time would kill node |
|
|
60 | connections in a bad way (patch by Malcolm Studd). |
|
|
61 | |
|
|
62 | 1.29 Fri May 7 20:13:39 CEST 2010 |
|
|
63 | - codename "Sadrak". |
|
|
64 | - fix error in callback at AnyEvent/MP/Global.pm line 339 |
|
|
65 | (found by Sadrak). |
|
|
66 | - listener-less nodes had trouble sending keepalive |
|
|
67 | messages on write timeouts (found by Sadrak). |
|
|
68 | - the monitor guard could cause a memleak due |
|
|
69 | to autovivification if a mon was cleared after |
|
|
70 | the port was gone (analysed by Sadrak). |
|
|
71 | - do not overwrite the config file if we couldn't read it |
|
|
72 | for some reason (lesson demonstrated by Sadrak). |
|
|
73 | |
|
|
74 | 1.28 Thu Apr 1 21:23:54 CEST 2010 |
|
|
75 | - accepted connections didn't correctly set up the SRCNODE |
|
|
76 | in some cases, leading to intra-node messages (such as monitoring) |
|
|
77 | to fail silently (great testcase by Sadrak). |
|
|
78 | |
|
|
79 | 1.27 Sat Mar 20 21:11:16 CET 2010 |
|
|
80 | - avoid exceptions in the transport code if the connection |
|
|
81 | fails during TLS negotiation. |
|
|
82 | - delay new address broadcasts in seednodes, to reduce the |
|
|
83 | likelyhood of nodes connecting at the same time, leading to |
|
|
84 | message loss. |
|
|
85 | - improve general message passing efficiency by 8%, and special |
|
|
86 | case json framing for a 50% speed increase. |
|
|
87 | - add hints how to interface to aemp nodes from other programming |
|
|
88 | languages using the aemp protocol in AnyEvent::MP::Transport |
|
|
89 | (it can really be quite simple). |
|
|
90 | - handle unsupported remote framings more gracefully. |
|
|
91 | |
|
|
92 | 1.26 Wed Dec 30 16:48:37 CET 2009 |
|
|
93 | - INCOMPATIBLE CHANGE: the high-level node protocol changed, |
|
|
94 | all nodes need to upgrade. |
|
|
95 | - remote kil's did not work (reported by Eugene Grigoriev). |
|
|
96 | - documented monitor_timeout, connect_interval, auth_offer, |
|
|
97 | auth_accept, framing_format, autocork and nodelay options |
|
|
98 | and make them configurable via bin/aemp. |
|
|
99 | - some doc updates. |
|
|
100 | - remote kil's were not synchronous to message sends which affected |
|
|
101 | nobody, as remote kils were broken. |
|
|
102 | - remote mon/unmonitor requests were not synchronous to message |
|
|
103 | sends either sometimes, although it likely didnt matter. |
|
|
104 | |
|
|
105 | 1.24 Wed Dec 9 14:59:52 CET 2009 |
|
|
106 | - codename "the foul egg". |
|
|
107 | - increased "unmonitored local port died" message level to 2, so |
|
|
108 | it is displayed by default now. |
|
|
109 | - generate random seed at configure time, not module loading time, |
|
|
110 | so fork&configure works as expected. |
|
|
111 | - decreased "seed"-related message priority in Global from 0 (d'oh) |
|
|
112 | to 9, as they are not signifcant. |
|
|
113 | - new experimental module AnyEvent::MP::DataConn to establish |
|
|
114 | additional data connections between nodes. |
|
|
115 | - bin/aemp now loads AnyEvent::Watchdog::Util before trying |
|
|
116 | to call restart. |
|
|
117 | - bin/aemp setcert didn't properly handle the filename argument. |
|
|
118 | - removed empty and unused timeout= parameter from protocol greeting. |
|
|
119 | |
|
|
120 | 1.23 Fri Nov 6 18:46:26 CET 2009 |
|
|
121 | - listener-less nodes were misinformed by their masters about |
|
|
122 | the location of new nodes, and therefore didn't connect |
|
|
123 | properly in all cases. |
|
|
124 | - messages send during node-up processing could sometimes get lost, |
|
|
125 | which would hamper global's ability to mesh the network. |
|
|
126 | - fixed AnyEvent::MP::Global::grp_reg to return a guard also |
|
|
127 | in scalar context. |
|
|
128 | - fixed AnyEvent::MP::Kernel::mon_nodes to return a guard also |
|
|
129 | in scalar context. |
|
|
130 | - try to improve error reporting when automatically loading |
|
|
131 | a function - being unable to load a module will now |
|
|
132 | stop the process and report the error. |
| 7 | |
133 | |
| 8 | 1.22 Sat Oct 17 03:41:47 CEST 2009 |
134 | 1.22 Sat Oct 17 03:41:47 CEST 2009 |
| 9 | - fix two rcv-bugs: after adding a tagged rcv, the default |
135 | - fix two rcv-bugs: after adding a tagged rcv, the default |
| 10 | callback would not be used or changed anymore (reported |
136 | callback would not be used or changed anymore (reported |
| 11 | by Felix Antonius Wilhelm Ostmann, with apologies for |
137 | by Felix Antonius Wilhelm Ostmann, with apologies for |
