Logs for jdev

Show join/part/nick changes:

[05:03:56] * ;;lk joined the chat.
[05:04:45] * ;;lk left the chat.
[05:37:46] * melomanozalo joined the chat.
[05:38:51] * melomanozalo left the chat.
[05:48:05] * teo left the chat.
[05:48:05] * teo joined the chat.
[05:56:58] * scippio left the chat.
[06:37:57] * Kev left the chat.
[06:44:15] * luca tagliaferri joined the chat.
[07:01:02] * ermine joined the chat.
[07:07:23] * luca tagliaferri left the chat.
[07:09:14] * smoku joined the chat.
[07:12:57] * smoku left the chat.
[07:15:59] * Guus joined the chat.
[07:53:47] * luca tagliaferri joined the chat.
[08:02:09] * waqas left the chat.
[08:02:10] * waqas joined the chat.
[08:05:43] * jprieur joined the chat.
[08:06:49] * jprieur left the chat.
[08:20:51] * Link Mauve joined the chat.
[08:37:28] * scippio joined the chat.
[08:56:50] * waqas left the chat.
[08:56:57] * waqas joined the chat.
[09:02:16] * MattJ joined the chat.
[09:05:49] * Neustradamus left the chat.
[09:10:19] * Treebilou left the chat.
[09:10:22] * Treebilou joined the chat.
[09:19:56] * waqas left the chat.
[09:27:01] * petermount joined the chat.
[09:49:20] * waqas joined the chat.
[10:08:08] * Kev joined the chat.
[10:19:20] * Treebilou left the chat.
[10:28:50] * Neustradamus joined the chat.
[10:32:17] * Treebilou joined the chat.
[10:33:20] * Treebilou left the chat.
[10:33:22] * Treebilou joined the chat.
[10:47:45] * alkino joined the chat.
[10:54:52] * rjhall joined the chat.
[10:55:47] <rjhall> Is there a part of the standards anywhere that describe how you might design your xmpp have two (sets of) XMPP servers servicing one domain?
[10:55:47] * bear left the chat.
[10:56:30] <Kev> rjhall: No, that's not really a protocol issue, it's an issue for the implementation of clustering in the server you use.
[10:56:50] <rjhall> so i have a bunch of users in the UK, and a bunch of users in the US. I don't want them to have different JID Domains (with federation) but the network between the two continents is flaky to the point where the people located where a single server isn't woul dbe unhappy
[10:58:33] <rjhall> Kev: that's what i thought... but thanks
[10:58:55] <MattJ> Yeah, it's quite tricky to do, and though I'd like to see a standard for it, I fully appreciate why there isn't one
[10:59:06] <Kev> I have run my domain like that, at one point - one node in the US, one node in the UK.
[10:59:27] <rjhall> right. i guess what i want is 'federation without a different domain' - which is kinda the whole antithesis of federation ;)
[10:59:37] <MattJ> It's called "clustering"
[10:59:45] <MattJ> Even if the cluster is rather wide :)
[10:59:58] <Zash> Dual!
[11:00:02] <rjhall> really? To me clustering is "lots of servers in one geographic location servicing a userbase"
[11:00:27] <MattJ> Yep, but a cluster serving one domain in XMPP may well be geographically dispersed :)
[11:01:30] <MattJ> That said, at least ejabberd doesn't do too well with nodes spread out like that
[11:02:08] <Kev> So I had heard.
[11:06:50] * mlundblad joined the chat.
[11:07:20] * mlundblad left the chat.
[11:09:17] <Guus> Don't bother trying Openfire for that either.
[11:09:23] <Guus> (if someone was planning)
[11:09:37] <Zash> MattJ: How's prosody clustering support comming?
[11:10:01] <Guus> although you might want to give Openfire's Connection Manager a try
[11:10:21] <MattJ> Guus, what's wrong with Openfire's in particular?
[11:10:30] <Guus> have one openfire server, use one or two connection managers, one of them located on the other continent
[11:10:39] <MattJ> Zash, currently 50 lines of code that don't do all that much :)
[11:10:44] <Guus> that will at least tunnel all of the socket connections over one socket
[11:11:26] <Guus> Openfire's clustering is basicaly not a lot more than replicating all caches, agressively. That'd give crap performance over slow lines
[11:11:50] <Guus> @MattJ: there are some synchronous calls over the cluster, which will effectively kill the domain that way.
[11:12:14] <MattJ> Zash, 1) I want to do it properly 2) It's going to take time 3) I can't afford that much time straight up, so I'm thinking of looking for sponsorship
[11:12:37] <MattJ> Guus, ok, gotcha
[11:12:48] <MattJ> Guus, sounds similar to ejabberd then
[11:12:55] <MattJ> except ejabberd is helped a little by erlang
[11:13:01] <Guus> could be - never looked at that.
[11:13:09] <MattJ> so you can't bring the server to a halt, but you can still mess things up
[11:13:13] * mlundblad joined the chat.
[11:13:35] <Guus> the synchronous calls are killing the Openfire cluster under stress - which is kind of defeating the purpose
[11:13:44] <MattJ> Yeah
[11:13:46] <Guus> you can run a cluster ok
[11:14:11] <Guus> but, where a single instance will run roughly 40k concurrent sessions, a cluster node will do just 10k
[11:14:26] <Guus> to give you an indication of the amount of overhead added.
[11:14:41] <MattJ> erk
[11:14:53] <Guus> but, @rjhall - you might want to give those connection managers a try
[11:15:10] <Guus> it's not clustering, but it'll allow you to aggregate all of the individual connections to your server
[11:15:13] <dwd> rjhall, FWIW, we refer to the scenario you're talking about as "wide area clustering".
[11:15:26] <Guus> that'll help - I'm not sure if it'll be enough for your purpose though.
[11:15:48] <dwd> rjhall, We did, originally, plan to do that over HF and things, but we changed our minds on that one - we now federate for that case.
[11:21:21] <MattJ> WAC... can we add a K to that somehow?
[11:21:48] <dwd> MattJ, I could ask Will?
[11:22:15] <MattJ> If you can organise a whitepaper on it, sure
[11:22:20] <MattJ> Lunch time -->
[11:46:34] <rjhall> dwd: HF?
[11:47:47] <Kev> Radio.
[11:53:52] * waqas left the chat.
[11:53:52] * waqas joined the chat.
[11:55:55] * mlundblad left the chat.
[11:56:21] * tkoski joined the chat.
[11:57:28] <dwd> rjhall, What's TVP, by the way?
[11:58:56] * mlundblad joined the chat.
[12:03:41] * teo left the chat.
[12:18:39] * mcepl joined the chat.
[12:18:40] * mcepl left the chat.
[12:24:18] * fantasticsid joined the chat.
[12:30:55] * benlangfeld joined the chat.
[12:32:46] * benlangfeld left the chat.
[12:35:22] <dwd> Well, that's odd.
[12:35:52] <dwd> Is there a SASL expert in the room? (Since neither of mine have got out of bed yet)?
[12:36:28] <petermount> i've barely touched sasl so I can't help
[12:37:02] <waqas> I wouldn't call me an expert, but what's odd? :)
[12:37:37] <dwd> XEP-0175 shows an example of an ANONYMOUS authentication.
[12:38:13] <dwd> Except no data is sent in the initial <auth/>, so shouldn't the server issue an empty <challenge/>?
[12:38:47] <MattJ> There's any data/challenge in ANONYMOUS?
[12:38:48] <dwd> I mean, shouldn't the client say <auth mechanism='ANONYMOUS'>=</auth>
[12:38:55] <MattJ> Ah
[12:38:59] <dwd> MattJ, Sure - look at RFC 4505.
[12:39:10] <MattJ> Yes, but iirc that's optional and not recommended
[12:39:18] <MattJ> (in XMPP)
[12:39:28] <dwd> MattJ, Right, but you say "=" to indicate you're explicitly not supplying anything.
[12:39:43] <MattJ> No, you say "=" to indicate you're sending an empty string
[12:39:56] <petermount> i'm just wondering about specific sasl implementations - i.e. for me Java has one and may expect it :-/ not sure off hand
[12:40:31] <MattJ> dwd, RFC: "The client may include in this message trace information [...]"
[12:40:40] <MattJ> So to not include, empty element
[12:40:50] <MattJ> If you want to include, but that information is "" then put "="
[12:40:59] <dwd> No.
[12:41:02] <MattJ> Yes.
[12:41:05] <dwd> Definitely not.
[12:41:15] <MattJ> Prove it.
[12:43:03] <petermount> hmmm, where's the "=" from?
[12:44:05] <dwd> RFC 4422: Where the mechanism is defined to allow the client to send data first, and the protocol's request message includes an optional initial response field, the client may include the response to the initial challenge in the authentication request message.
[12:45:02] <dwd> (In other words, the initial response is not mandatory - it's an optional field in the protocol's profile)
[12:45:34] <dwd> 3920bis: handshake for that particular authentication mechanism. This element MAY contain XML character data (in SASL terminology, the "initial response") if the mechanism supports or requires it; if the
[12:46:09] <dwd> In other words, the XML character data is the initial response, which, being optional, may be absent.
[12:46:25] <MattJ> Agreed
[12:46:48] <petermount> dwd: I think I see what you mean, yes it should be optional/absent
[12:47:26] * luca tagliaferri left the chat.
[12:47:42] <MattJ> dwd, but I don't conclude from this that the data being absent requires a challenge (this doesn't seem to be in the ANONYMOUS RFC)
[12:47:59] <dwd> MattJ, Erm - it's that way in the example.
[12:48:14] <dwd> MattJ, Since IMAP (base, at any rate) has no initial response.
[12:48:18] <MattJ> + means challenge in IMAP?
[12:48:28] <dwd> Yes.
[12:48:36] <MattJ> IMAP is broken you know
[12:48:42] <MattJ> But that's irrelevant :)
[12:48:58] <dwd> What's kind of annoying is that Gajim, at least, chokes when I send it a challenge.
[12:49:24] <dwd> MattJ, IMAP is perfect. I say this not just because Mark Crispin has more guns than me.
[12:49:29] <MattJ> Heh, I'd say that's a grey area
[12:49:39] <MattJ> + in IMAP is too ambiguous
[12:49:46] <MattJ> Its meaning depends on context
[12:50:10] <MattJ> and I don't like context in protocols
[12:50:53] <petermount> it complicates things too much ;-)
[12:50:54] <dwd> Not at all. + means "You haven't finished sending the command yet".
[12:51:14] <MattJ> Exactly, which doesn't mean "challenge" :)
[12:51:18] <dwd> The only interesting case is that in AUTHENTICATE, what the server says after the + is significant.
[12:51:39] <dwd> MattJ, That depends on your perspective. Given that SASL originated in IMAP, one might say it does by default.
[12:51:55] <MattJ> Heh
[12:54:43] <dwd> So, anyway, seems like ANONYMOUS has interop issues in XMPP.
[12:54:57] <MattJ> With one server? :)
[12:55:15] <MattJ> I do see where you're coming from now
[12:55:24] <dwd> MattJ, Well, I did actually write my own mechanism handler for it, but it's this way in Cyrus SASL, I think, too.
[12:55:50] <MattJ> Since a client that doesn't support initial responses will have no way to submit the trace data without a challenge
[12:55:50] * Alexander Gnauck joined the chat.
[12:56:00] * luca tagliaferri joined the chat.
[12:56:16] <dwd> Bollocks.
[12:56:28] <MattJ> and submitting the empty string is a decent way to say you've submitted the string, but have nothing to say in it
[12:56:37] * Alexander Gnauck left the chat.
[12:56:43] <dwd> MattJ, Not that that's legal in the spec either...
[12:56:53] <MattJ> Sending the empty string?
[12:56:55] * Alexander Gnauck joined the chat.
[12:57:11] * Alexander Gnauck left the chat.
[12:57:32] <dwd> MattJ, Seems that the message is mandated to be 1 octet long, minimum.
[12:57:47] <dwd> MattJ, token = 1*255TCHAR
[12:57:56] <dwd> (From RFC 4505)
[12:57:59] * niekie joined the chat.
[12:58:44] <dwd> Annoyingly, XEP-0175 passed Council while I was on it, and I got Kurt to review it carefully (as author of RFC 4505), so how this all got through I have no clue.
[12:58:54] <dwd> ... or any excuse.
[12:59:07] <MattJ> :)
[12:59:17] <MattJ> Yep, ok, I admit it - you're right
[12:59:35] <dwd> MattJ, Well, obviously.
[12:59:44] <waqas> Good thing I didn't say anything
[12:59:59] <MattJ> However the RFC author was evidently wrong to use "may" in that sentence
[13:00:07] <dwd> waqas, Allows you to step in and the end and day "Yeah!".
[13:00:13] <dwd> say, rather,
[13:00:19] <waqas> Yeah!
[13:09:55] * Guus left the chat.
[13:11:09] * Guus joined the chat.
[13:32:19] * niekie left the chat.
[13:39:46] <dwd> MattJ, BTW, what error do you give when you get a C2S stream to a MUC domain (for example)?
[13:40:14] <MattJ> Good question, I seem to remember we don't handle that as nicely as we should
[13:40:24] <MattJ> I think we just don't offer any way to authenticate
[13:41:03] <MattJ> In theory you could tell Prosody to load the SASL module for a component domain and log into components :)
[13:41:09] <dwd> Right - I think we offer authentication, but it'll never suceed.
[13:41:28] <MattJ> I don't think you'd receive any stanzas addressed to you though
[13:41:33] <dwd> Oh, indeed, if users at, say, conference.jabber.org existed, then Strange Things would happen.
[13:41:57] <MattJ> Well Prosody's MUC module allows you to set up a MUC on prosody@prosody.im, etc.
[13:42:08] <MattJ> but yes, strange things MAY happen :)
[13:42:12] * Tobias joined the chat.
[13:42:36] <dwd> Strange Things MAY happen. Administrators SHOULD NOT do this.
[13:42:52] <MattJ> Exactly
[13:43:12] <MattJ> Telepathy couldn't join such rooms because it checked the domainpart reported as a MUC service in disco
[13:43:28] <MattJ> But I hear they've removed that check now (because of G**gle I think)
[13:43:57] <waqas> I never could get a client to join a room without a localpart (that's supported too FWIW).. ^^
[13:45:20] <Zash> lolcatpart
[13:45:45] * niekie joined the chat.
[13:46:34] <waqas> I'm using a new keyboard, and am having to edit typos every few words.
[13:46:56] * lastsky left the chat.
[13:47:08] <Zash> /me is tierd and making up funny typos in his brain
[13:48:32] * fantasticsid left the chat.
[13:53:56] * Guus left the chat.
[13:56:14] * Guus joined the chat.
[14:05:06] * Zash left the chat.
[14:11:38] <dwd> MattJ, Oh, Kurt tells me I'm right about the XEP-0175 example being wrong, but the token is optional, so it can be missing.
[14:11:49] <dwd> MattJ, Which is to say, zero-length.
[14:12:04] <MattJ> missing != zero-length in my book
[14:13:07] <Tobias> yeah..zero length = ""
[14:14:16] <dwd> Tobias, Right, so it has to be supplied at some point.
[14:14:40] <dwd> MattJ, The token can be missing, the message cannot be, but id defined as [ addrspec / token ]
[14:14:55] <dwd> MattJ, ie, it can be "", an address, or a token.
[14:17:50] <rjhall> dwd: TVP is an office in Reading (UK) where i am now, rather than in California where i usually am. I set my online status in Psi for my corporate IM and it also set it here, i guess.
[14:18:21] <dwd> rjhall, Ah, gotcha.
[14:28:21] * niekie left the chat.
[14:44:52] * teo joined the chat.
[14:54:18] * stpeter joined the chat.
[15:03:06] * tofu left the chat.
[15:18:37] * tofu joined the chat.
[15:19:08] * alkino left the chat.
[15:21:13] * Zash joined the chat.
[15:24:15] * waqas left the chat.
[15:24:15] * waqas joined the chat.
[15:31:32] * waqas left the chat.
[15:31:49] * waqas joined the chat.
[15:32:23] * waqas left the chat.
[15:39:41] * hawke joined the chat.
[15:50:32] * Zash left the chat.
[16:02:29] * Kev left the chat.
[16:08:37] * bjc joined the chat.
[16:26:53] * Zash joined the chat.
[16:36:14] * waqas joined the chat.
[16:39:42] * jcea joined the chat.
[16:42:52] * evilotto joined the chat.
[16:52:25] * petermount left the chat.
[16:55:15] * tofu left the chat.
[16:56:58] * rjhall left the chat.
[16:58:06] * mlundblad_laptop joined the chat.
[17:01:34] * waqas left the chat.
[17:01:34] * waqas joined the chat.
[17:10:42] * scippio left the chat.
[17:19:57] * mlundblad_laptop left the chat.
[17:40:13] * luca tagliaferri left the chat.
[17:44:18] * tkoski left the chat.
[17:46:14] * tofu joined the chat.
[17:50:51] * MattJ left the chat.
[17:52:58] * evilotto left the chat.
[17:52:59] * evilotto joined the chat.
[17:53:33] * Tobias left the chat.
[18:04:28] * evilotto left the chat.
[18:05:27] * evilotto joined the chat.
[18:06:30] * Tobias joined the chat.
[18:12:41] * smoku joined the chat.
[18:13:28] * evilotto left the chat.
[18:13:53] * evilotto joined the chat.
[18:14:58] * evilotto left the chat.
[18:15:03] * evilotto joined the chat.
[18:19:34] * Zash left the chat.
[18:25:06] * waqas left the chat.
[18:25:15] * tkoski joined the chat.
[18:36:41] * teo left the chat.
[18:36:42] * teo joined the chat.
[18:44:27] * jprieur joined the chat.
[18:44:31] * jprieur left the chat.
[18:45:23] * jprieur joined the chat.
[18:45:25] * jprieur left the chat.
[18:47:53] * Florob joined the chat.
[18:56:00] * nabatt left the chat.
[19:29:58] * scippio joined the chat.
[19:42:24] * Zash joined the chat.
[20:04:18] * smoku left the chat.
[20:05:36] * tkoski left the chat.
[20:35:45] * Neustradamus joined the chat.
[20:35:49] * mlundblad joined the chat.
[20:35:58] * jcea joined the chat.
[20:36:33] * elmex joined the chat.
[20:37:11] * dwd joined the chat.
[20:37:19] * evilotto joined the chat.
[20:37:51] * smoku joined the chat.
[20:37:51] * smoku left the chat.
[20:38:19] * Guus joined the chat.
[20:38:21] * Neustradamus left the chat.
[20:38:29] * scippio joined the chat.
[20:39:26] * Neustradamus joined the chat.
[20:41:43] * stpeter joined the chat.
[20:44:24] * Zash joined the chat.
[20:46:46] * zanchin joined the chat.
[20:50:46] * tofu joined the chat.
[20:51:33] * Tobias joined the chat.
[21:17:31] * tofu left the chat.
[21:26:01] * Treebilou joined the chat.
[21:27:26] * Treebilou left the chat.
[21:31:54] * Treebilou joined the chat.
[21:42:26] * mlundblad left the chat.
[21:45:56] * Tobias left the chat.
[21:46:56] * Treebilou left the chat.
[21:54:01] * Zash left the chat.
[21:54:06] * Florob joined the chat.
[21:54:13] * Zash joined the chat.
[21:59:50] * rjhall joined the chat.
[22:09:57] * rjhall left the chat.
[22:11:08] * Zash left the chat.
[22:11:49] * Zash joined the chat.
[22:13:07] * scippio left the chat.
[22:14:47] * tofu joined the chat.
[22:20:16] * johnny joined the chat.
[22:28:20] * Zash left the chat.
[22:28:38] * Zash joined the chat.
[22:43:07] * Zash left the chat.
[22:43:09] * Zash joined the chat.
[22:47:31] * hawke joined the chat.
[22:52:27] * stpeter left the chat.
[23:00:57] * Guus left the chat.
[23:01:30] * Florob left the chat.
[23:13:21] * hawke left the chat.
[23:20:25] * hawke joined the chat.
[23:28:40] * hawke left the chat.
[23:34:05] * Zash left the chat.
[00:59:58] * evilotto left the chat.
[01:26:28] * jcea left the chat.
[01:38:55] * luca tagliaferri joined the chat.
[02:17:09] * Zash joined the chat.
[02:57:26] * Zash left the chat.
[03:05:12] * fantasticsid joined the chat.
[03:16:19] * lastsky joined the chat.
[04:11:36] * waqas joined the chat.
[04:35:46] * waqas left the chat.
[04:35:46] * waqas joined the chat.