#ossasepia Logs for 03 Oct 2019



April 20th, 2020 by Diana Coman
diana_coman: !o uptime [03:33]
ossabot: diana_coman: time since my last reconnect : 0d 23h 27m [03:33]
diana_coman: !o help [06:53]
ossabot: diana_coman: my valid commands are: src, uptime, seen-anywhere, help, s, version, seen [06:53]
diana_coman: !o version [06:53]
ossabot: I am bot version 597604. [06:53]
auctionbot: B#1061 O=161mn LB=151mn E=2019-10-03 05:55:34.297573 (5h55) >>> 1209 Wired Filthy Fiats, WU is fine but not preferred [08:49]
auctionbot: --- end of auction list, 151mn total bids --- [08:49]
diana_coman: thimbronion: how did you order that list of irc networks? the top100 from netsplit.de seems different (possibly changed since you looked but seems rather unlikely to have changed that much) [09:55]
thimbronion: diana_coman I've been working on them from top to bottom according to the source, but I haven't explicitly tried to keep them in any order. Last night I was considering at least putting them in order of number of servers, with those having linking applications available at the the top. [11:04]
diana_coman: thimbronion: makes sense. [12:02]
diana_coman: fwiw that oftc thing seems quite approachable, from among the "big" ones [12:02]
thimbronion: I recall Undernet seeming approachable as well. EsperNet also looked good, but it is on the smaller end at 12 servers. [12:14]
diana_coman: thimbronion: you need number of servers vs users I'd say ie number of servers by itself is not telling much [12:17]
diana_coman: maybe they need more servers because they have too few or something [12:17]
thimbronion: diana_coman: Ah yes that makes sense. [12:18]
diana_coman: thimbronion: so maybe order the list based on what you think ie "seems most likely to accept my server" [12:18]
diana_coman: once you have at least a few that seem on first pass as a possible option, we can go in and talk to them. [12:19]
thimbronion: diana_coman: so should I finish researching the remaining 65 or do an ordering now and start talking to people? [12:21]
thimbronion: or both... [12:21]
thimbronion: It's quite an archeological dig. Most website findings are at the mid 2000s layer. [12:22]
diana_coman: thimbronion: if you have already a few (2-4 at least) that seem in the least approachable, just connect to them first and look around and then start talking to them; let me know what/which ones you picked and I can come in too if it helps [13:19]
diana_coman: yeah, looking at the list does reveal lots of "dead but listed" pretty much [13:19]
asciilifeform: diana_coman: fixed today ugly bugola in reader.py ( thrd here ) [13:22]
snsabot: (trilema) 2019-10-03 asciilifeform: ACHTUNG, PANZERS! logotron www updated ! [13:22]
diana_coman: asciilifeform: aite, I'll manually change it for now. [13:23]
asciilifeform: ty diana_coman [13:24]
asciilifeform: i really must get rid of the positional print statements in that thing [13:25]
asciilifeform: they're nuffin but trouble. ( instead string concats like normal people ) [13:25]
diana_coman: should be on, let me know if you see any weird [13:26]
asciilifeform: diana_coman: loox proper. [13:26]
asciilifeform: only lobbes nao remains w/ the barf, i expect he'll fix when wakes up. [13:28]
diana_coman: thimbronion: re continuing the list - in principle it's worth doing anyway because you'll need it later in any case; but it can be done at a slower pace/don't need to wait for it to be complete before starting talking to them [13:28]
diana_coman: thimbronion: and where "archaeological", write it as a note "not updated since 2012" or whatever [13:28]
diana_coman: the point is: document the thing; here we looked at all sorts and this is what was found. [13:29]
asciilifeform: diana_coman: 1 moar thing, i have an autosyncer, but wrote it while ill, so to avoid similar headaches will post it 1ce fully alive [13:31]
asciilifeform: ( semi-auto, to be pedantic. takes a cutoff point; drops errything in all chans after that point; then walks list of e.g. http://logs.nosuchlabs.com , http://logs.ossasepia.com , fetches errything in all chans from each, then eats ~the longest~ ) [13:33]
thimbronion: diana_coman: Ok. Perhaps it makes sense add some documentation discussing the general quality of the results at the beginning of the post and also add a "notes" line to each entry? [13:33]
asciilifeform: diana_coman: this aint ideal -- i can picture cases where it cannot be used , if multiple bots dropped at various times -- but beats the shit outta manual munging [13:34]
diana_coman: thimbronion: sounds good, yes. [13:42]
diana_coman: asciilifeform: sure, I'll have a look when you post it but take your time. [13:43]
diana_coman: asciilifeform: re longest, I'd say it should prolly look at a diff or similar though because in some cases one might need a mixture of them, ugh [13:44]
diana_coman: tbh at resync time the usual headache is when/if lines are missed in the middle ie need to drop some & then update; but anyways, as starter it's better than nothing, for sure [13:46]
asciilifeform: diana_coman: this is correct, and in fact is moar serious, from mathematical pov, problem , than appears at 1st glance [16:01]
asciilifeform: diana_coman: my draft semi-auto syncer simply asks for a tuple chan,line that represents the point of break; then takes its timestamp, drops ~everything~ in db that postdates that time; then proceeds to algo described earlier. [16:05]
asciilifeform: ( there is no attempt to mathematically derive the 'holes', it will work strictly if one of the N given peers contains the complete log from T to present time ) [16:06]
asciilifeform: if the proggy were built into bot, there would -- in theory -- be no need to do the drop, it could run as part of the reconnector sequence. however it is impossible to rule out the possibility that a line is spoken ~while~ the syncer syncs. therefore i'ma leave it as a manual crank for the foreseeable future. [16:07]
asciilifeform: fwiw i suspect that this mechanism will be obsolete when we get sumthing like a irc net going. [16:07]
asciilifeform: diana_coman: there is 1 potential point of danger re the described syncer -- if the 'break line' chosen was ~itself~ imported from another bot, it may result in unwanted droppage ( there is no guarantee of monotonicity of timestamps except as written by ~one particular~ bot, atm ) [16:09]
asciilifeform: i'm thinking of having it display the lines to be dropped, and prompt, prior to firing, atm cannot think of a cleaner pill. [16:10]
diana_coman: if anyone wants to follow this particular thread, the discussion of proposed sync is in #trilema as it belongs better there. [16:30]
ossabot: (trilema) 2019-10-03 asciilifeform: folx w/ mathematical inclinations, invited to comment re auto-sync algo described in #o . [16:30]
snsabot: Logged on 2019-10-03 15:46:07 asciilifeform: diana_coman: this is correct, and in fact is moar serious, from mathematical pov, problem , than appears at 1st glance [16:30]
thimbronion: diana_coman: when it comes to speaking with server admins on IRC, I am not sure what approach to take. From looking at most applications, linkers appear to be in the position of having to sell themselves, and I admit that sales is a field in which I have much to learn. [16:41]
thimbronion: diana_coman: alternatively, I could just approach it from an information gathering perspective. "Hi, is there any interest in new servers, what are the chances of the application being accepted, etc." [16:42]
diana_coman: thimbronion: we are not really selling anything given that they aren't paying, you know? [16:42]
diana_coman: we are offering them a chance; they are free to be idiots and miss out on it, just like freenode did [16:43]
diana_coman: thimbronion: but yes, I know you aren't yet clear on this part yet [16:43]
thimbronion: diana_coman: this is true. And this being true, I find their attitude perplexing. [16:43]
diana_coman: thimbronion: and no, don't approach them begging/probing because you won't be able to get out of that initial framing [16:44]
diana_coman: and it's not helping [16:44]
diana_coman: thimbronion: their attitude is their problem, really; it's perplexing only if you look at it with some pre-conceived expectations really [16:44]
diana_coman: so don't; you focus on what *you* offer and otherwise let the rest of the world figure out for themselves what they are going to do about it and how they are going to handle it [16:45]
diana_coman: if they handle it idiotically fine, it's their loss anyway. [16:45]
thimbronion: Many, many Bob Becks. [16:45]
diana_coman: plus you get clear proof that they are idiots, what. [16:46]
thimbronion: diana_coman: May I convey myself as representing tbf? [16:47]
diana_coman: thimbronion: let me know what networks you choose, maybe dig out the old freenode-managed-to-reject-server thing too , for some idea [16:47]
diana_coman: thimbronion: tbf is currently on a rather unclear footing really, not sure how that helps; you *are* representing TMSR [16:48]
diana_coman: thimbronion: however, before asking them re federating server, the first step is to just join some chans and look around, read for a bit [16:50]
thimbronion: diana_coman: Got it. The few chans I've joined so far have been almost completely dead for days. [16:51]
diana_coman: thimbronion: I suspect that's extremely true for...most of them, myeah [16:51]
diana_coman: what networks did you join? [16:52]
thimbronion: IRCNet and p2p-net. [16:52]
diana_coman: thimbronion: looking at your list, I see undernet, dalnet, ircnet, oftc, europnet maybe [16:56]
diana_coman: iirc there was efnet mentioned too, though I don't see it in your list [16:56]
diana_coman: europnet supposedly aims specifically to "integrate across europe" so it would be particularly lulzy if they ...refuse to integrate [16:57]
diana_coman: thimbronion: what software are you going to run? how well do you know it? do you have any experience with it? [16:57]
thimbronion: diana_coman: yes I need to re-add and document EFNet. I beleive I couldn't find the linking application or linking was discouraged, but as I said I will add it in with documentation. [16:59]
thimbronion: diana_coman: which software? IRC client or IRC server? [16:59]
diana_coman: thimbronion: the server; you'll need to run it and in time bridge it to other networks supposedly, no? [17:00]
thimbronion: diana_coman: I haven't made that determination. I ran my own ircd years ago, but I don't recall which and I was pretty much the only person using it. So my experience is almost nil. The quickest (though perhaps not wisest) approach would be to run the server of the first network that allows us to link. [17:04]
thimbronion: Then later select perhaps the most robust irc server (what the criteria for that would be I don't yet know) and start that running and then build the bridge from the network's custom ircd to our own. [17:05]
diana_coman: thimbronion: the thing is that you are in a much better position to talk to them if you already actually have your server basically [17:05]
diana_coman: ie I have this server and chans and people, active unlike your shit, do you federate my server or should I find others? [17:06]
thimbronion: diana_coman: I recall many applications requiring that you run their own custom server. [17:07]
diana_coman: thimbronion: and given how ~dead they seem in general + what passes for "awesome software solutions" nowadays, I half-expect "their custom ircd" to not even run on a sane tmsr environment [17:07]
diana_coman: thimbronion: that's still rather 2nd consideration as far as I see it; ie you already have something and are reaching out; they can then require all sorts and you'll consider [17:08]
thimbronion: diana_coman: related: I was able to build DALNet's ircd, but haven't run it yet. [17:08]
thimbronion: diana_coman: also many networks want you to run ntpd [17:09]
thimbronion: speaking of sanity [17:10]
diana_coman: thimbronion: the point is: don't let apparent requests deter you upfront anyway; but to be able to talk to them from a proper position, you do need to know very well what you are running there and what you are willing to do or not; but let *them* ask if they want some nonsense, anyway [17:12]
diana_coman: thimbronion: so why not run that dalnet ircd anyway? [17:12]
thimbronion: diana_coman: it has the size of dalnet going for it. [17:12]
diana_coman: for that matter not even sure - how would they exactly check that you are running "their ircd"? [17:12]
diana_coman: thimbronion: you know, the trouble with "size of dalnet" is pretty much the trouble with "1mn unique visitors" on blogs [17:13]
diana_coman: ie sure, numbers; whether they add up however to any real activity is still dubious [17:13]
thimbronion: diana_coman: according to the applications, there is a 1 week or whatever link evaluation period [17:13]
thimbronion: diana_coman: I don't know what kind of evaluating happens though [17:14]
diana_coman: thimbronion: eh, so let them evaluate and if anything speak up, etc [17:14]
diana_coman: anyway, not yet there [17:14]
diana_coman: the point is: you *will* need to get to know that software, whatever you pick [17:14]
diana_coman: so if you want to pick dalnet's, on whatever ground, fine; if you want to pick something else, fine; [17:15]
diana_coman: but in any case, you'll need to *bridge* it [17:15]
diana_coman: so avoiding this in the 1st step is not really earning you anything [17:15]
thimbronion: diana_coman: So let's say I build the bridge first. How do I test it "in wartime conditions"? I guess I would need to run my own copy of whatever other network's ircd somewhere else and test against it, and perhaps write some bots to do stress testing. [17:17]
diana_coman: thimbronion: first you pick and stand up some ircd; then you go as representative of tmsr and armed with all the logs re activity + chans + users and talk to networks, see which one is saner; they can come with all sorts of requests but you can consider them first, it's not like you have to do anything they ask [17:20]
thimbronion: diana_coman: ok, understood. [17:22]
diana_coman: thimbronion: re testing in wartime conditions - this is valid for anything anyway ie if you want to do proper testing upfront, you still need the whole orchestra, no/ [17:22]
diana_coman: alternatively, you can say well, there's no wartime-testing better than the war itself and plunge in but no idea how hairy it can/may get; I don't really have experience running ircd [17:23]
diana_coman: thimbronion: does that make sense? is there some bit/part that you'd rather approach differently (how)? [17:24]
diana_coman: thimbronion: for that matter re testing, after standing up ircd, you mirror the tmsr chans here with the help of existing loggers for instance [17:26]
thimbronion: diana_coman: when you say "logs re activity + chans + users" are the the logs etc of the targeted network, or our own? [17:27]
diana_coman: thimbronion: our own; I doubt they even have logs, do they? [17:28]
thimbronion: diana_coman: initially I though I was to log some of their most active channels, etc. to have own record of the size/activity on the selected network. [17:29]
diana_coman: but after all, the situation is precisely this: we want to move away from freenode and we have the resources to stand up our own server; we think network X is sensible and won't turn down some additional resources, so how about they link our server ? [17:30]
diana_coman: thimbronion: ah, no; I doubt that's any use really [17:30]
diana_coman: you know, that list of requirements and so on is mostly for you to do an initial filtering of the most obviously-idiotic; but otherwise let *them* point you at it and request whatever; you have the stuff and make them an offer, that's about it [17:32]
diana_coman: and obv, you'll blog the result either way. [17:32]
diana_coman: thimbronion: does this clear up a bit the "not sales" part at least [17:33]
diana_coman: ? [17:33]
thimbronion: diana_coman: Sorry for the delay, trying to restate in own words. [17:36]
thimbronion: diana_coman: actully, nm: http://ossasepia.com/2020/04/20/ossasepia-logs-for-03-Oct-2019#1004538 sums it up quite well. [17:37]
ossabot: Logged on 2019-10-03 17:30:35 diana_coman: but after all, the situation is precisely this: we want to move away from freenode and we have the resources to stand up our own server; we think network X is sensible and won't turn down some additional resources, so how about they link our server ? [17:37]
thimbronion: diana_coman: so yes, this makes sense. [17:38]
diana_coman: thimbronion: good; ask/speak up if there is/appears later anything else that doesn't make full sense to you because you do need to have a very clear idea of it before you can successfully talk about it, ofc [17:39]
thimbronion: diana_coman: ok, I will. [17:40]
diana_coman: thimbronion: re ircd, it's really up to you re version or whatever but honestly, I'd rather choose the smallest/most minimal thing to start with, just to be able to figure it out [17:40]
diana_coman: iirc it's supposed to be modular so in principle you can add to it as needed? [17:41]
diana_coman: doesn't know ircd in any real detail [17:41]
thimbronion: diana_coman: I, too, do not yet know ircd or the protocol itself in much detail. Small minimal makes sense from that perspective. If it turns out it's missing something every network seems to need, we can re-evaluate or add to the bridge. [17:47]
diana_coman: thimbronion: dunno if that irc services thing is of any use/interest/maybe help still; e.g. achurch.org/services/docs-5.1.24/tech/5.html [17:49]
thimbronion: diana_coman: noted, thankyou. I will review it. [17:52]
asciilifeform: http://logs.nosuchlabs.com/log/ossasepia/2019-10-03#1004552 << last time i picked up the subj (ftr broke teeth) found that most (all?) of the major nets use some nonstandard 'embrace&extend' crock of shit, rather than irc per strict rfc-specsheet. (which is not to say that cannot be sewn w/ skilled hands. simply, may not suffice to 'just pick a ircd' naively) [19:21]
snsabot: Logged on 2019-10-03 17:25:49 diana_coman: doesn't know ircd in any real detail [19:21]
asciilifeform: in some cases may be possible to 'cheat' and stick to rfc-traditional protocol. ( as i did in re client vs fleanode. i refused to use fleanode's proprietary knob for confirming success of joining chans, instead simply inserted delay ) [19:22]
asciilifeform: ( represented by 'join_t' knob in bot config ) [19:23]
asciilifeform: fwiw tho, mp's orig. proposal was to obtain whatever ircd was distributed by ea. heathen net, and patch'em to interoperate together after. [19:26]
asciilifeform: ( if any of these prove egregiously broken, can simply omit that one ) [19:28]
snsabot: Logged on 2019-10-03 16:51:36 diana_coman: thimbronion: and given how ~dead they seem in general + what passes for "awesome software solutions" nowadays, I half-expect "their custom ircd" to not even run on a sane tmsr environment [19:28]
BingoBoingo: Deedbot hasn't given me a challenge to decrypt for voice yet. Datacenter has been contacted [22:38]
asciilifeform: all chans synced as of nao. [22:39]
thimbronion: http://ossasepia.com/2020/04/20/ossasepia-logs-for-03-Oct-2019#1004511 << Correction - I built Undernet's ircd. [23:25]
ossabot: Logged on 2019-10-03 17:08:53 thimbronion: diana_coman: related: I was able to build DALNet's ircd, but haven't run it yet. [23:25]

Comments feed: RSS 2.0

Leave a Reply