IRC log for #infobot on 20080509

06:52.04troubledFactBot: dvd playback troubleshooting is <reply> testing
06:52.04FactBottroubled: okay
06:52.09troubled$dvd playback troubleshooting
06:52.09FactBottesting
06:52.13troubled$forget dvd playback troubleshooting
06:52.13FactBoti forgot dvd playback troubleshooting, troubled
06:52.38troubleddondelelcaro: re: #debian-bots, any chance you could look at the db key for that entry and see what might be causing it to choke?
06:53.17troubledmy bot seems fine, not sure why dpkg wouldnt be able to find it yet show it in a literal query. perhaps the logs showing something?
06:53.24troubledanyways, awaiting response. l8r
10:24.21*** join/#infobot Breakz (n=Obliviou@3E339CE5.dslaccess.aol.com)
11:06.14*** join/#infobot Breakz (n=Obliviou@3E339CE5.dslaccess.aol.com)
11:06.40Breakzi must say, the old infobot was much nicer even though more annoying
11:07.13Breakzat least it rlly memorised stuff and responded without having to do crap like ~ or search for what he replies too
11:09.45troubledBreakz: the bot has different modes it can run it dont forget
11:10.55Breakzi like this bot though, but it misses what i actually loved from the old infobot
11:11.03Breakzthe annoyance
11:11.04Breakz:P
11:11.22troubledheh
11:11.30troubledhavent used the old bot yet myself
11:11.32Breakzthat also if ur normally talking it replies random crap for no reason ;p,  ut that also made it handy
11:11.40troubledhow does it differ thats so anoying?
11:11.41Breakznow its ~forums to see our forum link
11:11.50Breakzwhere the old replied to Forums? where are the forums? omg forums?
11:11.52Breakzand other combo's
11:12.08troubledyou can set it to non addressed mode for that i think
11:12.17Breakzi've set it to non addressed mode
11:12.19troubledand you can change the ~ to something else btw
11:12.23Breakzand to the highest level of annoying and w/e
11:12.32Breakzdoesnt change
11:12.58troubledcan you paste your config with any nick/pass info censored?
11:14.15Breakz# [str] addressing is when you name the bot. FIXME:
11:14.15Breakz#   REQUIRE     -- the bot only does something if addressed.
11:14.15Breakz#   OPTIONAL    -- the bot responds (does not learn) irrelevent of
11:14.16Breakz#                  addressing.
11:14.16Breakzset addressing  OPTIONAL
11:16.00troubledthere is another setting or two that might be affecting that
11:16.21troubled# [str] when should the bot bother learning new factoids.
11:16.23troubled#   ADDRESSED   -- only learn when addressed.
11:16.25troubled#   HUNGRY      -- learn irrelevent of addressing. this will catch
11:16.28troubled#                  _everything_, use at your own risk.
11:16.30troubledset learn               ADDRESSED
11:16.33troubledthats one
11:16.38troubled# [str] different behaviour with URLs.
11:16.40troubled#   REQUIRE     -- means it will need to be a url type (e.g. file:, http:)
11:16.43troubled#   OPTIONAL    -- will take anything
11:16.45troubled#   REJECT      -- will not accept any urls.  this makes it easy to
11:16.47troubled#                  run 2 with different nicks and styles.
11:16.50troubled#                  ^^^ what's the point of this???
11:16.52troubledset acceptUrl           OPTIONAL
11:16.54troubledthats the other
11:17.23troubledI think if its in OPTIONAL mode, you need to set the acceptUrl to OPTIONAL as well, else it may only pick up http://..../ urls with exact syntax
11:21.33Breakz# [str] different behaviour with URLs.
11:21.34Breakz#   REQUIRE     -- means it will need to be a url type (e.g. file:, http:)
11:21.34Breakz#   OPTIONAL    -- will take anything
11:21.34Breakz#   REJECT      -- will not accept any urls.  this makes it easy to
11:21.34Breakz#                  run 2 with different nicks and styles.
11:21.34Breakz#                  ^^^ what's the point of this???
11:21.36Breakzset acceptUrl           OPTIONAL
11:21.38Breakzgot that
11:23.39troubledso you have OPTIONAL, HUNGRY and OPTIONAL for those 3?
11:24.01Breakzyup
11:24.02Breakz# [str] when should the bot bother learning new factoids.
11:24.02Breakz#   ADDRESSED   -- only learn when addressed.
11:24.02Breakz#   HUNGRY      -- learn irrelevent of addressing. this will catch
11:24.02Breakz#                  _everything_, use at your own risk.
11:24.02Breakzset learn        HUNGRY
11:24.16Breakzsame as the old infobot actually
11:24.26Breakzyet i need to address it with `
11:24.27Breakz~
11:24.43troubledso in your channel, when you say something like: foobar is available for download at http://foobar.com/, does it remember it?
11:24.48Breakzyea
11:24.51troubledthe ~ is changed with chanset
11:25.04troubledlike: botname: chanset #thischan addressCharacter !
11:25.04Breakzi dont want to address it
11:25.10Breakzi want it to jibberjabber random shit
11:25.13Breakzlike the old infobot
11:25.40troubledif you did that foobar example, what happens if you do: foobar?
11:25.48Breakzit looks like the best part of the old infobot got jacked out of this one
11:25.53Breakzno then u gotta do ~foobar
11:26.02Breakzit doesnt reply to anything like 'foobar?'
11:26.12Breakzsince it NEEDS to be addressed at all times no matter what config ur using
11:26.17troubledI think the problem is that ? used to force a question is now used to force a bot check for missing factoids. could be a regex issue stopping execution from continuing to get what you want
11:27.00troubledBreakz: try that "foobar?" example though, curious if that works
11:27.16troubled(after you set foobar as a factoid of course)
11:28.03Breakz<05*09::13*27*34> <~Shadow> foobar is available for download at http://foobar.com/
11:28.03Breakz<05*09::13*27*35> <~Shadow> foobar?
11:28.04Breakz<05*09::13*27*37> <~Shadow> ~foobar
11:28.04Breakz<05*09::13*27*38> <+Ometotchtli> foobar is, like, available for download at http://foobar.com/
11:28.34troubledyou want it to just work for "foobar?" only then I assume?
11:28.46Breakzyea like the old
11:28.48Breakzfoobar?
11:28.51Breakzwhere can i find foobar/
11:28.55Breakzwhat is foobar?
11:28.56Breakzetc.
11:29.14troubled"what is foobar?" doesnt work either i take it?
11:29.35troubledmind me, but ive seen the regex section where all the hooks for that stuff are. its crazy :)
11:29.35Breakznope
11:29.44Breakzand for that im almost switching back to the old infobot
11:30.06Breakzwhere this one has much more features and the features work better then old infobots features
11:30.15troubledwell, something like that could probably be added, but it might need to be discussed with the other dev guys to work out specifics
11:30.24Breakzadded?
11:30.29Breakzit was in infobot :p
11:30.34troubledkeep in mind that such a change would really hammer your db
11:30.35Breakzits ripped out :x
11:30.38troubledheh
11:30.44Breakzidc, its what infobot made infobot
11:30.46troubledwas before my time unfortunately
11:30.53Breakzwithout that
11:31.01Breakzits basicly... a normal average ircbot
11:31.25troubledpersonally, i think that syntax should be allowed with the addressed mode being optional
11:31.46troubledi do know though that we use the ? to force inter bot checks for unknown factoids though
11:31.50Breakzyea that addressed mode setting actually works and does something
11:32.50troubledfor example, if I had FactBot linked to infobot here and FactBot didnt know "foobar", but infobot did, I could do: FactBot: foobar?, and the ? would force it to check the other bots it knows. that could be where it broke, but it would have to be checked thoroughly to know for sure
11:33.25Breakziunno, im a noob at perl :P
11:33.30troubledsame here hehe
11:33.35Breakzim still trying to convince my own DEVS
11:33.38troubledcan you believe they gave me commit access? ;)
11:33.39Breakzto add a AFK funtion
11:33.45Breakzlike: ~afk gone
11:33.58troubledwhat should that do?
11:33.59BreakzBreakz? it replies> is afk reason <gone> for <time left>
11:34.18Breakzlots of ppl dont got afk stuff in scripts or if ur somewhat important
11:34.22Breakzits handy that a bot does that
11:34.31troubledmost people dont expect a channel bot to go afk I guess
11:34.56Breakznot the bot
11:34.57Breakzthe person
11:35.14troubledthe devs that run these bots in big channels tend to only run them one way, and very open. so sometimes certain bugs slip past everyone for old features that nobody knows or cares about
11:35.16Breakzi've got this function in my old bot to, but i'd like it in infobot more
11:36.00Breakz~AFK im away for holiday
11:36.02Breakzbreakz?
11:36.11troubledah
11:36.17BreakzBreakz is AFK Reason: im away for holiday, been gone for 3min
11:36.19Breakzlike that
11:36.46troubledmost people would probably use or suggest !seen for that stuff
11:36.56Breakzdifferent
11:37.11troubledlong as the last thing you said was "AFK, out of the country", that would show up fine in seen
11:37.13Breakzi've had it in my bot
11:37.18Breakzand almost everyone who knew it, loved it
11:37.43Breakzyea perhaps, butif ur Afk and u say sumtin or w/e :p
11:37.59Breakzlike if ur still busy but talking to DEVs or w/e in another channel ;x
11:38.04troubledah
11:38.21troubledi could see !afk being useful. although that would be another reserved word gone heh
11:38.44Breakzyea perhaps, but i think its a handy feature
11:39.05Breakznow im using: ~AFK Breakz is <reply> AFK: Gone for bla
11:39.06Breakz:p
11:40.20troubledlol
11:40.30troubledrube goldberg for sure :)
11:40.38Breakz?
11:40.53troubledinfobot: rube goldberg
11:40.53infobotmethinks rube goldberg is a cartoonist who created crazy complicated contraptions for performing various mundane tasks; see http://www.rube-goldberg.com/html/gallery.htm
11:41.23troubledyou know, those stupid machines that go all around the house setting stuff off to do something simple like squeeze orange juice
11:42.49Breakznah never heard of it :p
11:42.59troubledsurely youve seen them before though
11:43.22troubledhttp://video.google.com/videoplay?docid=-2367646121273499414
11:43.26Breakzperhaps
11:43.26troubledthere is an example of one
11:43.34Breakzi stopped watching internet humour stuff quite a while ago
11:43.47troubledbasically its the hard way to accomplish something simple
11:43.58troubledanyways heh
11:45.04Breakzdude must be so fuking bored
11:45.06troubledas for your afk idea, I might be able to make that for ya, but i would need to talk to the other guys about specifics and get an okay on a design
11:45.14Breakzcool
11:45.35troubledfile storage would be fine, but I think were really trying to get moved over to full db rather than away from it heh
11:45.36Breakzi wish i was a coder lol
11:45.54Breakzmeh, the db is small
11:45.57troubledso schema has to be discussed, as well as the changed to the main regex stuff that does all the checks for the command and stuff
11:46.17Breakzcompared to my other db's, infobots db traffic is lower then my forums db traffic lol
11:46.34troubledwell, _your_ db is small, but these guys run 50mb+ db's. so a query or two extra may not mean much to you, but it could really hurt the bigger infobots
11:46.56Breakzheh its about 10mb atm
11:47.02troubledi still think they got plenty of breathing room until they hit a few million, but still like to keep things efficient heh
11:47.20Breakzthe compile the perl code to c
11:47.44Breakzcould speed up lots of stuff
11:48.00troubledas for the other issue, I cant see it being a problem to allow the "<factoid>?" syntax check for a factoid if the bot is in optional addressed mode. shouldnt affect the bigger bots
11:48.14troubledyou think the bot is slow?! :)
11:48.32Breakzits as fast as my mirc coded bot
11:48.35Breakz;)
11:48.44troubledI know I can get like 10k reads/sec on this old machine with dbm which isnt much off from sqlite
11:49.26troubledill admit that the bot is a pig though. I think they run like 18-20mb typical on fresh install startups which needs to be addressed. hoepfully POE would solve that if we ever get there :)
11:49.57Breakzzomg
11:50.04BreakzPOE::CGI::IRC?
11:50.11troublednot cgi
11:50.18Breakzbetter not
11:50.22Breakzits much slower then sockets
11:50.46troubledPoCi isnt it?
11:50.53troubledbeen a while since I read up on all that poe stuff
11:51.00Breakzi've used the poe cgi stuff
11:51.21Breakzits slower as mirc lol
11:51.35troubledi tend to shy away from cgi myself. might make for simpler pipes between modules or something, but i just dont like it for speed for some reason
11:51.38troubledseems inefficient
11:51.39Breakzperl is like 6-7 times slower then C, mirc is about 20 times slower then C lol XD
11:51.55Breakzand cgi is even slower as mirc
11:52.18Breakzplus they added PHP like noob protection innit
11:52.31Breakzdont fuckingadd gay PHP stuff into PERL modules :'(
11:53.45troubledheheh
11:54.37troubledwell, we have talked about a rough direction to head with the bot. a bit of a cleanup and POE were definetly well accepted, but generally on the condition that things like config syntax stay stable and unchanged as much as possible
11:55.05troubledie: if conf changes and syntax get a make over, it should be _after_ a 100% comptible POE rewrite, not before or during
11:55.08Breakzyea and add back the old infobot annoying stuff
11:55.11Breakzoptional or not ;p
11:55.22Breakzjust like 'set imannoying <true/false>'
11:55.23Breakz:p
11:55.33troubledi would actually have to use the old bot and see. but I think POE would be a better effort myself
11:56.31Breakztry the old bot and i hope u can see what i mean :P
11:56.44Breakzit would respond mid-convo because it sees a fact :p
11:57.05Breakzquite annoying at first, but on my server we started to love it after a while XD
11:57.55troubledwell, i cant see don or tim using it since the channels they are in would fall into chaos if the bot did that with the factoids they have hehe
11:58.11troubledbut if it were linked to that addressed mode specifically, it should be good
11:58.13Breakzyea but thats why, make it optional
11:58.38Breakzjust like the addressed was for.. where now doent matter what option u set it :P it just responds to addressing anyways
11:58.48troubledit probably broke along the way some time and nobody noticed since these guys dont use that feature. sadly, we are in dire need of some regression tests for that problem
11:59.28Breakzwell i can test stuff it needed
11:59.43Breakzgot some perl coders who could test functions/exploits etc.
11:59.48troubledlet me take a look at code for the bot real quick for the foobar? stuff
12:00.11troubledfeel free, patches are always welcome, as are bug reports and feature requests via sourceforge
12:00.37troubledsec, let me loginto my lil' test net and file up a bot with that setting and see what goes on internally
12:02.25Breakzmeh, i'm running a few months old infobot/blootbot thing so might be updated or not, iunno
12:02.43troubledyou able to join a test server?
12:04.06Breakzim running a 'test' server
12:04.07Breakzlol
12:04.11Breakzdebian lenny/sid
12:04.12BreakzXD
12:05.00troubled:)
12:05.40Breakzi use it as a normal public though, not recommed by debian ppl :P but who cares i have more package then any other debian version and there often the most updated :P
12:06.18troubledya, all debian here too
12:06.31Breakzyea noticed with the inbuild package checker :p
12:06.41Breakzanyways
12:06.52Breakzits sunny outside, im gonna chill out with some mates in the park for a few hours
12:06.53Breakzbbl
12:07.06Breakzconnected with -j #channel, so eventually i'd be joining there
12:07.08Breakzttyl man :)
12:07.27troublednp, cya around
14:52.05*** join/#infobot TimRiker (n=timr@68.27.76.234)
14:52.06*** mode/#infobot [+o TimRiker] by ChanServ
14:56.08troubledTimRiker: hey you
14:56.43troubledthis Breakz guy had a few requests if you check infobot logs for the last hour or two to catch up. but i want to ask you about the patch I got so far
14:57.05troubledinvolves hacking up a bunch of broken regex in the Core.pl and Question.pl files
14:57.31troubledill pastebin a rough uncleaned patch in a bit for you to mull over if you got time
15:02.54troubledTimRiker: http://pastebin.ca/1012754
15:03.45troubledneeds a cleanup and such, but commented out line in question.pl is the big one. Seems that Breakz was upset that the old infobot way of replying to "question?" was broken
15:04.55troubledthat minVolunteer check was always failing for one, and second I have no clue wtf its for yet without going through svn history  i suppose. But i did get that query? format working on my test bot so far. Need to test when its not in hungry mode with optional addressing though just to be sure
15:05.22troubleddondelelcaro: feel free to look at that url i pasted to tim. it expires in 1 day
15:06.58troubledone of the problems though is that the "is" part was always being left in, so i just s/// the extra info instead of changing stuff to /where is/ which seems to work. Seems like a proper place to do it as well since otherwise we would have to put duplicate logic in Question.pl to strip out the "where is" and all that crap.
15:08.49troubledalso, wth is with all this \Q's without matching \E's? I added the two I saw, but im not entirely sure if they need the enclosing \E if they arent using any other matching in the regex. feedback would be appreciated on what the right thing to do is
15:09.27troubledon a side note, I think im about to have a mental breakdown after having seen the Core.pl code :(
15:09.34troubledwere in trouble heh
15:09.52troubledwords cant describe the amount of hackery going on in there
15:10.52troubledanyways, look over the patch, put your feedback and changes/patches in here and ill get back to you to discuss. also, silence = commit, so speak up before hand ;)
15:27.56TimRikerreads
15:32.33TimRiker$rhs =~ /^\Q$lhs /i and last;  <- you removed a needed space, no?
15:33.15troubledseems like it. only meant to add \E though. ill adjust in final patch I guess
15:33.18TimRikerwhy did you kill the "hey, " >
15:33.21TimRikerwhy did you kill the "hey, " ?
15:33.25troubleddidnt work :/
15:33.32troubledsame with "stupid question: "
15:33.45troubledwell, stupid question works 100%, but "question:" doesnt
15:33.46TimRikerinfobot: hey, where is bzflag?
15:33.49infobotTimRiker: what are you talking about?
15:33.54TimRikerinfobot: where is bzflag?
15:33.56infobotI think you lost me on that one, TimRiker
15:34.00TimRikerinfobot: where is bzflag
15:34.01infobotTimRiker: what are you talking about?
15:34.13TimRikerlame bot
15:34.18troubledany regex with a token using (  )? seems to be failing
15:34.55troubledi got that patch on a test bot on my test net atm and it seems to be running as expected with that patch
15:35.27TimRikerinfobot: bzflag
15:35.28infobotbzflag is probably a 3D internet multiplayer multiplatform (win32, linux, mac, etc) opensource opengl Battlezone capture the flag game of the same name that you must try at http://BZFlag.org/ or a continual development project with periodic gaming interuptions. See also TimRiker
15:35.30troubledalthough im testing optional addressed modes with non url checks, so its a tad yappy. but the Question.pl patch did seem to unbreak the "question?" stuff
15:35.31TimRikerinfobot: bzflag?
15:35.31infoboti heard bzflag is a 3D internet multiplayer multiplatform (win32, linux, mac, etc) opensource opengl Battlezone capture the flag game of the same name that you must try at http://BZFlag.org/ or a continual development project with periodic gaming interuptions. See also TimRiker
15:35.38TimRikerinfobot: where is bzflag?
15:35.39infobotI think you lost me on that one, TimRiker
15:35.41TimRikerinfobot: where is bzflag ?
15:35.43infobotTimRiker: I think you lost me on that one
15:35.50TimRikerstupid bot
15:36.12troubledjoin infobot.doesntexist.org port 9997 again if you like
15:36.21troubledhanging in #infobot of course
15:36.47troubledBreakz is over there too atm since this is a feature request of sorts for him
15:37.13troubledwith that \Q stuff though, isnt the \E mandatory closing?
15:37.19TimRikersure. looks like cleaning it up and testing it is a Good Thing. Should document in there some examples of what should work.
15:37.31TimRikerdunno bout the \E.
15:37.55troubledwell, theres two of the \E stuff, and im not sure if space in regex should be " " or "\ "
15:38.24troubledother than that, i was just gonna test some of the regex, add/change a few and play around to see what breaks with all the diff addres modes
15:39.57troubledaslong as addressed only mode doesnt break, it should be good. but I still need to check into why stuff that doesnt end in a ? gets looked up in the db twice for no reason. seems like it could be painful in lots of big channels in non addressed mode (ya, like we'll every try that heh)
15:39.59TimRikerjust space.
15:40.19troubled10:43 <@TheStaff> that and I need to look into why its pulling in anything like that hehe
15:40.22troubled10:43 < infobot> DEBUG: question = that and I need to look into why its pulling in anything like that hehe
15:40.26troubled10:43 < infobot> DEBUG: question = that and I need to look into why its pulling in anything like that hehe
15:40.49troubledcurrently, that patch is causing that ^^^, double db queries. might be long standing bug though, have to look
15:41.00TimRikerdunno
15:41.18troubledsucks tracking this stuff down you know, the logic is all over the place heh
15:42.02TimRikerdefault minVolunteerLength is 50. not that I'm sure what that means.
15:42.15troubledi have nfc
15:42.33troubledi grepped for it, aside from the conf, its only appearing in the Question.pl and in some other file
15:42.49troubledwhatever it was for though, broke "query?"'s in Question.pl though
15:43.24troubledi could see maybe using it as a max length for the query to avoid looking up stupidly long question, but it should be max not min then
15:43.39troubledso if the length($query) > 50, return, kinda deal
15:44.06troubledthat would work and make sense to me, but i have no idea what "min" is meant to do otherwise
15:44.42TimRikerlocal $finalQMark = $query =~ s/\?+\s*$//;     should NOT change query.
15:44.52troubledi could probably look it up in svn, but with all the perltidy commits, I know its gonna be a pita since ill have to pin branches all the way back to where it was inserted
15:45.32troubledthats just to strip out the ? it seems
15:45.44TimRikerthat does not strip anything.
15:46.13TimRikerat least I don't think it does.
15:46.13troubledzero or one "?" followed by any number of space til EOL, gets erased
15:46.34troubledso: "?                " -> ""
15:46.39TimRikerI think it'll just set finalQMark to true if it matches. perhaps it does do a change too. I'd have to try a test script.
15:46.58TimRikerI've normally only used that syntax with m/
15:47.00troubledthats probably fine, but we need to strip out that ? somewhere in tehre
15:47.51TimRikerprobably should handle "?+ *"
15:48.06TimRikeroh, there is a plus in there. duh
15:48.17troubledthat or strip out the ? in the caller instead and just have doQuestion take the actual factoid to look up
15:48.56troubledthis is why i hate patching this bot. if you want to truely fix one thing, you gotta redesign half a module to do it heh
15:49.05TimRikerwell, the issue is that there if you have  ? in the channel, you may want to answer. where as with addressed, it makes no difference.
15:50.49troubledbut should doQuestion() really do anything more than just lookup the param passed in the db?
15:51.36troubledi guess i am proposing that the logic be moved back to Core.pl for now until it can be decided to be split up elsewhere if/when we do a cleanup on the structure
15:51.39TimRikersome one centrally needs to decide if we are going to do a lookup and answer.
15:52.04TimRikercore is a bad place for it, as other modules call question now. ...
15:52.17troubledhmmm, depending on how many spots in Core.pl, that could add 4 or 5 diff repeated chunks of code to do the logic
15:52.21TimRikerprobably want a maybequestion or something as well as a doquestion
15:52.50TimRikerso that it can be explicit if someone else is just passing on and letting us decide, or actually demanding a reply.
15:52.57troubledi dont mind if the sub/logic is in the Question.pl, but if it could be removed from doQuestion() so its leaner and self explanitory, it might help
15:53.14TimRikerfor that matter, question is a bad name now, as some things are commands and not questions at all. :)
15:53.27TimRikerit's all in how far you want to dig.
15:53.27troubledtalk about a can of worms heh
15:54.00TimRikerso if we're not addressed, and there is no question mark we do nothing..
15:54.31troubledi think someone mentioned a problem with that in the code
15:54.38TimRikerif there is a question mark and minVolunteerLength is 0 or unset, we do nothing.
15:54.44troubledmore like a complaint that the check for that shouldnt be so far down the code
15:55.08troublednamely the :INFOBOT stuff is waaaay down there
15:55.21TimRikerif the length of the line is < 50 (after editing) and there was a question mark, we reply
15:55.26TimRikeror try to.
15:55.31troubledseems like that should just be moved to a diff sub in that file and called as "return thatsub()"
15:55.58troubledwhat does "volunteer" mean in that case though
15:58.28troubledas I said, I could understand if it was like "$maxLengthOfQueryBeforeIgnore" or something. then anything past 50 could be ignored question wise
15:58.29TimRiker:INFOBOT:QUERY <timriker> bzflag
15:58.39troubledthink its msg hook only
15:58.46TimRiker:INFOBOT:QUERY <timriker> bzflag?
15:58.49TimRikerchecking.
15:59.27TimRikeryep. msg only and works as expected.
15:59.34troubledpossible blootbot guy coming from #debian-bots here btw
16:00.05TimRikerhuh?
16:00.06troubledim surprised that interbot code still works :)
16:00.20TimRiker:)
16:00.27troubledjust some guy called yang in #debian-bots asking about blootbot, told him to c'mon over here
16:00.34TimRikerah.
16:00.41troublednm, simonrvn gabbin at him
16:01.02TimRikerlistens
16:01.04troubledyou know, you need to get a headset and an IAX/SIP softphone :p
16:01.32troublednot sure what your take on voice communication is though. irc people tend to be picky about voice these days
16:01.36TimRikerheh. nah. I'm always in multiple conversations. irc handles that, voice does not scale.
16:02.05*** join/#infobot yang (i=yang@static-ip-62-75-255-125.inaddr.intergenia.de)
16:02.08yanghello
16:02.13simonrvnhi
16:02.13troublednot for large stuff, but was thinkin might be great to verbally discuss how to edit some of these files
16:02.16troubledhey
16:02.36troubledgotta love when we get cross channel like this :)
16:02.58troubledTimRiker: ok, far as branch goes, you are using trunk or a tag?
16:03.25troubledkinda hoping to get all but the test bots onto tags so we can freely break trunk for faster tests and cleanup for a bit
16:03.52troubledmaybe get the topic changed to tell new users to grab the latest tag instead of trunk as well to avoid hassles
16:04.32TimRikerhelp should change to reflect addressing. ie: 'I learn mainly by observing declarative statements such as "x is at http://www.xxx.com"' should say something like "you can teach me by saying ..." or "if someone in the channel says ..."
16:04.36troublednot that I intend to break trunk if I can help it of course, but it does seem easier to just commit rapidly to trunk instead of doing the whole branch, edit, merge process
16:04.49troubledya, i noticed that
16:05.17TimRikertroubled: trunk. I update locally and test a bit, then update the bot.
16:05.48troubledwould require a heavy hack in the help code though. Maybe some sort of $ADDRESSED=OPTIONAL$ syntax in the help to allow for the check for var's or somethin
16:06.11troubledwell, long as you keep an eye open for trunk patches then :)
16:06.20TimRikernods
16:07.45troubledTimRiker: ok, so \E after \Q should be safe provided I fix the spaces I removed?
16:07.59TimRikeras far as I can see, yes.
16:08.00troubledjust wanna clean up that patch and commit
16:08.05troubledk brb
16:08.31TimRikerbut killing the "return '' if ( length $query < &::getChanConf('minVolunteerLength') );" does not look right to me.
16:09.01troubledgrep for that minVol var in the whole tree, i dont see it actually used anywhere useful
16:09.24troubledaside from that file, its only in 1 other code module
16:09.33TimRikerdon't nuke things just cause you don't see how they are used.
16:10.00TimRikersomeone added it for some purpose. find out what, test it, and fix it. don't just nuke it without understanding.
16:10.28troubledk, let me do digging the svn logs and see if i can find that commit line
16:10.42TimRikerwe are not documented well enough to handle things that way. more documentation in the code would be a Good Thing. :)
16:10.57TimRikeras to your "why is this defined again?"
16:11.42troubledi think its safe, but let me present my case first :)
16:11.55TimRikerI think that the ? check should be: $finalQMark += $query =~ s/\s*\?\s*$//; or something like that.
16:12.14TimRikerI think it's trying to handle "what is foo ? ? ? ? ?" and doing it poorly.
16:12.33troubledgrep -Ri minVolunteerLength * | grep -v ".svn"
16:12.35troubledfiles/infobot.chan~:    minVolunteerLength 50
16:12.37troubledfiles/infobot.chan:    minVolunteerLength 50
16:12.39troubledfiles/sample/infobot.chan:    minVolunteerLength 50
16:12.41troubledsrc/IRC/IrcHelpers.pl:    if ( &IsChanConf('minVolunteerLength') > 0 ) {
16:12.44troubledsrc/Factoids/Question.pl:        return '' unless &IsChanConf('minVolunteerLength') > 0;
16:12.47troubledsrc/Factoids/Question.pl:#        return '' if ( length $query < &::getChanConf('minVolunteerLength') );
16:12.50troubledsrc/Factoids/Question.pl:#       What is minVolunteerLength? Enabling it breaks original infobot
16:13.14TimRikershould handle "foo?", "foo ?", "foo ? ", "foo??????? ", "foo ????????", "foo ? ? ? " etc.
16:13.15troubledso beside that file, its only in IrcHelpers.pl, and i wasnt going to nuke that check
16:13.35TimRikeryeah, I've seen all those. I grepped the source too.
16:13.40troubledTimRiker: ya, i think it does mostly. the Core.pl has the logic to fix duplicates and spaces and stuff
16:14.11troubledso as far as Question.pl is concerned, its a beningn chunk of code
16:14.26troubledlet me digg up the commit for that line too though so we can at least see whats up
16:15.20TimRikerwhat is bzflag?
16:15.43troubled<PROTECTED>
16:15.48TimRikerinfobot: chanset #infobot minVolunteerLength 3
16:15.48infobotSetting minVolunteerLength for #infobot to '3'.
16:15.51TimRikerwhat is bzflag?
16:15.52infobotACTION is bzflag to understand recursion, you must first understand recursion
16:15.52troubledwuzthat! ;)
16:16.08TimRikerthat was odd.
16:16.14TimRikerwhat is literal bzflag?
16:16.15infobotACTION is literal bzflag to understand recursion, you must first understand recursion
16:16.25TimRikerliteral bzflag?
16:16.25infobot"bzflag" is "a 3D internet multiplayer multiplatform (win32, linux, mac, etc) opensource opengl Battlezone capture the flag game of the same name that you must try at http://BZFlag.org/ or a continual development project with periodic gaming interuptions. See also TimRiker"
16:16.29TimRikerbzflag?
16:16.30infobotbzflag is a 3D internet multiplayer multiplatform (win32, linux, mac, etc) opensource opengl Battlezone capture the flag game of the same name that you must try at http://BZFlag.org/ or a continual development project with periodic gaming interuptions. See also TimRiker
16:16.39troubledr1147 | timriker | 2005-02-18 02:51:53 -0500 (Fri, 18 Feb 2005) | 2 lines
16:16.42troubledminVolunteerLength now per channel
16:16.49troubledstill going back though, sec
16:16.54TimRikerah. so that does not help you much. :)
16:17.19TimRikercmd: what (.*?)?
16:17.25TimRikercmd: what is (.*?)?
16:17.34troublednah, looking at 0r1347 as well for the line above it. see if that helps
16:17.42troubledr1347 | timriker | 2007-05-21 02:56:45 -0400 (Mon, 21 May 2007) | 1 line
16:17.44troubledtake a few more things literally
16:17.49troublednor does that heh
16:18.28TimRiker:)
16:18.35TimRikerbzflag?
16:18.35infobotbzflag is a 3D internet multiplayer multiplatform (win32, linux, mac, etc) opensource opengl Battlezone capture the flag game of the same name that you must try at http://BZFlag.org/ or a continual development project with periodic gaming interuptions. See also TimRiker
16:18.41TimRikerwhat is bzflag?
16:18.41infobotACTION is bzflag to understand recursion, you must first understand recursion
16:18.46TimRikerwhere is bzflag?
16:19.06TimRikerinfobot: listkeys cmd: what
16:19.27TimRikerinfobot: literal cmd: what (.*?)
16:19.28infobot"cmd: what (.*?)" is "($1): <action> $1 to understand recursion, you must first understand recursion"
16:19.47TimRikerinfobot: factinfo cmd: what (.*?)
16:19.47infobotcmd: what (.*?) -- created by raventh1 <n=raven@68.142.142.101> at Wed Jul 18 19:56:08 2007 (295 days); it has been requested 257 times, last by TimRiker, 1m 6s ago.
16:20.06TimRikerannoying misfeature.
16:20.13TimRikerinfobot: forget cmd: what (.*?)
16:20.13infobotTimRiker: i forgot cmd: what (.*?)
16:20.19TimRikerwhat is bzflag?
16:20.27TimRikerit should answer to that one.
16:20.41TimRikerso minVolunteerLength is working as designed.
16:20.54TimRikerwhere is bzflag?
16:21.01TimRikerbut that one is broken.
16:21.05TimRikerwhere bzflag?
16:21.06infobotbzflag is a 3D internet multiplayer multiplatform (win32, linux, mac, etc) opensource opengl Battlezone capture the flag game of the same name that you must try at http://BZFlag.org/ or a continual development project with periodic gaming interuptions. See also TimRiker
16:21.11TimRikerbzflag?
16:21.11infobotbzflag is a 3D internet multiplayer multiplatform (win32, linux, mac, etc) opensource opengl Battlezone capture the flag game of the same name that you must try at http://BZFlag.org/ or a continual development project with periodic gaming interuptions. See also TimRiker
16:21.16TimRikerthose work.
16:21.23troubledthat with that patch?
16:21.28TimRikerno.
16:21.32TimRikerthat's before the patch.
16:21.58troubledthe real testing needs to be done in hungry/nonaddressed mode though
16:22.23TimRikerI was just testing nonaddressed mode. seems to work ok.
16:22.49troubledwhat about just: "query?"
16:22.54TimRikerbzflag?
16:22.54infobotbzflag is a 3D internet multiplayer multiplatform (win32, linux, mac, etc) opensource opengl Battlezone capture the flag game of the same name that you must try at http://BZFlag.org/ or a continual development project with periodic gaming interuptions. See also TimRiker
16:22.59TimRikerlike that?
16:23.00infobotACTION smooches that on the cheek
16:23.04TimRikerhehe
16:23.08troubledis bzflag a module commandstub hook?
16:23.14TimRikerno
16:23.19troubledjust a factoid?
16:23.24TimRikerbzflist?
16:23.35TimRikerinfobot: bzflist
16:23.46troubleddebian?
16:23.46infobotsomebody said debian was the hobby project of bruce perens that flew on the space shuttle
16:23.58TimRikercommand works in addressing. only factoid works unaddressed.
16:24.12TimRikerinfobot: literal debian
16:24.13infobot"debian" is "the hobby project of bruce perens that flew on the space shuttle"
16:24.20TimRikerinfobot: factinfo debian
16:24.20infobotdebian -- created by atoponce <n=aaron@ubuntu/member/pdpc.supporter.bronze.atoponce> at Fri Sep  7 01:57:55 2007 (245 days); it has been requested 17 times, last by troubled, 34s ago.
16:24.37troubledand you are running trunk too :/
16:24.58troubledwonder why it was broke with the settings I tested then
16:25.00TimRikerinfobot: debian is also http://www.debian.org.  See http://www.debian.org/doc/manuals/project-history/index.en.html#contents
16:25.01infobotACTION tells is also http://www.debian.org.  See http://www.debian.org/doc/manuals/project-history/index.en.html#contents to RTFM!!!! GAH!!! HELL FIRE AND BRIMSTONE!!!! BURN!!! DIE!!! MWAHAHAHAHA!!!!
16:25.13TimRikerhuh?
16:25.14infobotWelcome to ICQ.
16:25.19troubledI know for sure that the ? was dying on that line I was going to nuke
16:25.35simonrvnwtf wtf
16:25.56troubledim using: set learn               HUNGRY
16:26.00troubledset acceptUrl           OPTIONAL
16:26.14troubledset addressing          OPTIONAL
16:26.33troubledset those 3 and try
16:26.49troubledcause "query?" was for sure broken on trunk on the bot i tested on
16:27.55TimRikerit's fine here. please don't break it. :)
16:28.44troubledi think the !$addressed is whats allowing yours to work
16:28.57troubled$reload
16:28.57FactBotreloading...
16:28.58FactBotreloaded:
16:29.04TimRikerperhaps
16:29.06troubledblender?
16:29.07infobot[blender] Extremely fast and versatile 3D Rendering Package. URL: http://freshmeat.net/projects/blender/
16:29.13TimRikerinfobot: blender?
16:29.14infobotblender is Extremely fast and versatile 3D Rendering Package. URL: http://freshmeat.net/projects/blender/
16:29.19TimRikerinfobot: where blender?
16:29.20infobotblender is Extremely fast and versatile 3D Rendering Package. URL: http://freshmeat.net/projects/blender/
16:29.23troubleda reload should work for incofbot.config?
16:29.25TimRikerinfobot: where blender
16:29.25infobotblender is Extremely fast and versatile 3D Rendering Package. URL: http://freshmeat.net/projects/blender/
16:29.29TimRikerinfobot: blender
16:29.30infobotblender is Extremely fast and versatile 3D Rendering Package. URL: http://freshmeat.net/projects/blender/
16:29.34TimRikerall work just fine.
16:29.52TimRikerNO. reload does not reload the config afaik.
16:29.58troubledthats what I though
16:30.01troubledsec then
16:31.37TimRikerinfobot: no, debian is http://www.debian.org.  See http://www.debian.org/doc/manuals/project-history/index.en.html#contents or the hobby project of bruce perens that flew on the space shuttle
16:31.43TimRikersilly bot.
16:31.52troubledok, bringing factbot back with hungry/optional/optional
16:31.54*** join/#infobot FactBot (n=factbot@pdpc/supporter/sustaining/troubled)
16:31.54*** mode/#infobot [+v FactBot] by ChanServ
16:31.59troubledthanks
16:32.00infobottroubled: bitte
16:32.00FactBottroubled: no problem
16:32.06troubledok, that passed :)
16:32.14troubledton?
16:32.19troubled$ton
16:32.19FactBotTon Roosendaal is the lead developer of the free 3D application Blender, and chairman of the Blender Foundation.
16:32.22troubledthats broken
16:32.28troubledwho is ton?
16:32.34TimRikerwhat's min on yours?
16:32.41troubled50(default)
16:32.48troubledfactbot is using trunk
16:33.03TimRikerinfobot: chanset #infobot minVolunteerLength
16:33.03infobotminVolunteerLength for #infobot is '3'
16:33.09troubledmy other testbot is using the patch. i know its $addressed related, ive seen the line that does the check
16:33.25TimRikerso you will ignore "ton?"
16:33.26troubled$chanset minVolunteerLength
16:33.26FactBotShowing minVolunteerLength values on all channels...
16:33.27FactBot<PROTECTED>
16:33.27FactBot<PROTECTED>
16:33.28FactBotEnd of list.
16:33.56TimRikerset minVolunteerLength to '1' if you want it to always reply here.
16:34.07troubledthe patch allows "ton?" to work becuase when I put in the debug statements, its quiting the if check on that minVolunteerLength line
16:34.26TimRikeryour minVolunteerLength is incorrect. the code is fine.
16:34.35troubledthat could be a good hack/feature, but i really think we should at least add something more descriptive
16:34.47troubledits a useless check though :/
16:35.00TimRikerno! it's a useful check as I just demonstrated.
16:35.20TimRikerleave it there. it works _exactly_ as designed.
16:35.32troubledterrible name then :p
16:35.55troubled$chanset #infobot minVolunteerLength 1
16:35.56FactBotSetting minVolunteerLength for #infobot to '1'.
16:35.58troubledton?
16:35.59FactBotTon Roosendaal is the lead developer of the free 3D application Blender, and chairman of the Blender Foundation.
16:36.19TimRikerit's the minimum length of a line that I should process to see if I want to volunteer a reply. that sounds exactly like what the name means.
16:36.21troubledim fine having it as a setting, but I see a few problems
16:36.54troubled1, the default is bad in that hungry mode is broken unless you grep src code, and second the name could be better. perhaps a new config entry with a better name?
16:37.28troubledI guess that would be a good comment for it
16:37.28TimRikerperhaps we should remove addressing completely and point the docs to minVolunteerLength
16:38.04TimRikerdefault minVolunteerLength to unset in _default and something small in #infobot (or #botpark, or some other silly place)
16:38.06troubleddont you think its backwards?
16:38.17troubledjust seems like it should go opposite and use  "max" instead
16:38.23TimRikernot at all. what are you after?
16:38.36TimRikerI can see that a max would _also_ make sense.
16:38.36troubledthen we can say "anything past 50 chars, dont even bother checking for a question"
16:39.14troubledthis way when I type: Hi, blah blah blah......................200chars later...so anyone help?, it doesnt hit the db with a 400 char query
16:39.48TimRikergood point that. /me ponders.
16:39.52troubledlets be realistic, its max would be like the 32 char (forget var) key max + the length off the longest "where is" or "who is" stuff
16:40.40TimRikerthe length is calculated _after_ the where is, who is, etc is removed.
16:40.55troubledah, nm in that case then
16:41.18TimRikerso perhaps just a return '' if $length ? maxkeylen  would make sense?
16:41.26TimRikerer > maxkeylen
16:41.38troubledso how about switching min for max of 50, and change min to something sane like 3+?
16:42.16TimRikerand killing "addressing" ?
16:42.31troubledno, was thinking that needs to stay
16:42.41troubledelse you would piss don off in #debian when the bot goes haywire :)
16:44.03TimRikerI don't really see the point in having both min/max _and_ addressing.
16:44.33TimRikerif min/max is unset, or some other non-working value, then I'm in address-only mode. this should be the default.
16:44.34troubledwell, min is to save from useless queries, and falls in line with the bot flag needed to search <= 2 char factoids
16:44.49troubledthe max is to eliminate silly long messages that clearly dont need a db lookup
16:44.53TimRikerthen I can set min/max in one channel if I like, on the fly, and there is no need to restart the bot.
16:45.11troubledso if I paste a 450 char message, the bot should clearly _not_ look that up and just ignore it
16:45.20troubledi like that idea as well
16:45.28TimRikeragreed. but that does not justify keeping addressing.
16:45.43troubledpicture #debian without it then. lets mock up a situation
16:45.52TimRikerwithout what?
16:45.58troubledwithout addressing
16:46.10troubledlets say its non hungry mode for this setup
16:46.17TimRikerwithout the code having a valiable called addressing? or with addressing optional?
16:46.21troubledsomeone in debian says: hello?
16:46.47troubledim assuming complete removal unless you mean otherwise
16:47.09TimRikerthe bot sees that min/max volunteer length means it should ignore the "hello" and so it does.
16:47.10troubledthe issue I see is that dpkg in #debian would start repsonding fairly often to random chatter
16:47.55TimRikerno. the default setting should be to to require addressing. ie: min/max set to something that gets that behavior.
16:47.56troubledbut if you set the min to a point that it avoids hello, then you cut out a large chunk of the factoids it can do and could confuse the users using it
16:48.14troubledits impossible to get that behaviour from min/max alone though
16:48.21TimRikerthe min/max is for _volunteer_ replies. which you don't want.
16:48.30TimRikerie has no bearing on direct questions.
16:48.53TimRikerthat's what $addressed handles.
16:49.01troubledwhat i am thinking is to use min/max as a control of when to check the db. min to limit :)'s from hitting the db (minor), but namely max needed to make long msg's not hit the db either
16:49.41TimRikerall db ?s ? or just the volunteer ?s as it is supposed to me?
16:49.45TimRikers/me/be/
16:50.19troubledso a volunteer question is like "ton?" or "who is foobar?"?
16:50.38TimRikerright. and not ~ton or "infobot: ton" etc.
16:50.53troubledwhich are addressed, but you want to remove in this scenario right?
16:50.59TimRikerNO!
16:51.02troubledoh
16:51.23TimRikerI want to remove the bot-wide addressing option. NOT the $addressed variable.
16:51.23troubledwhen you said remove $addressed, what did you mean then?
16:51.28troubledahh
16:51.45troubledso you want it to be per channel instead? or not even an option?
16:51.47TimRikerremove addressing. not $addressed.
16:52.48TimRikerI want to use the min/max volunteer length instead of "set addressing"
16:53.30TimRikerfor that matter, I'd like to nuke the bot-wide "set learn" and have a per channel flag for that.
16:53.32troubleddefinetly would work. im just wondering about how we would deal with users wanting the old functionality like don
16:53.51troubledarguably, half the config should be per channel :)
16:54.19TimRikerthe default should still be require addressing. so min/max could be unset, or set to some thing like -1/-1 or just to 100000/1 etc.
16:54.38TimRikeras much of the config as we can make per-channel is a Good Thing.
16:54.53TimRikerI've converted a ton of stuff over the years. like addressing character, etc.
16:54.56troublednow, far min = 1 or so, do you not think that sorta conflicts with the bot flag to restrict 2char or less searches in the db?
16:55.24troubledjust seems wrong that we have a flag to query them, but could slip by and allow 2 char lookups if min is set to a default of 1 or 2
16:55.31TimRikerit should check both now, right? if we pass on and have a 1 char query, we should not end up looking it up anyway.
16:55.57troubledwould have to see where that check is i guess and make sure min/max cant bypass that check
16:56.07TimRikerso you need to get by _both_ limits. as well as the max key length.
16:56.23TimRikerright. I don't expect they bypass it now.
16:56.24troubledbesides, I think that 2char check is only for listkeys and listvalue stuff, so I guess its irrelevant
16:56.27TimRikerbut worth checking.
16:56.34TimRikercould be.
16:56.40TimRikerinfobot: a
16:56.40infobota is not b  The first letter in the English alphabet!
16:56.54TimRikerone characters questions work. :)
16:57.07TimRikerinfobot: listkeys a
16:57.13TimRikernada there.
16:57.22troubledyour admin though, let me try
16:57.25TimRikeroh... too soon. :)
16:57.31troubledinfobot: listkeys b
16:57.39troubled<infobot> search string is too short.
16:57.45troubledso ya, only for searches
16:58.10TimRiker"a" works for you?
16:58.18troubledwell, tried b, but no
16:58.21troubledi need +o iirc
16:58.24TimRikernot a search...
16:58.25troubled$+help flags
16:58.26FactBot<PROTECTED>
16:58.26FactBot<PROTECTED>
16:58.27FactBot<PROTECTED>
16:58.27FactBot<PROTECTED>
16:58.28FactBot<PROTECTED>
16:58.28FactBot<PROTECTED>
16:58.29FactBot<PROTECTED>
16:58.29TimRikerjust a question.
16:58.31FactBot<PROTECTED>
16:58.31troubledoh, sec
16:58.33FactBot<PROTECTED>
16:58.34troubledinfobot: b
16:58.34infobotpicobot: c
16:58.35FactBot<PROTECTED>
16:58.37FactBot<PROTECTED>
16:58.39FactBot<PROTECTED>
16:58.41FactBot<PROTECTED>
16:58.43FactBot<PROTECTED>
16:58.45FactBot<PROTECTED>
16:58.54TimRikerheh. #botpark loop factoid.
16:59.00TimRiker~factinfo b
16:59.00infobotb -- created by fragglet <~fraggle@pc2-glfd1-4-cust151.gfd.cable.ntl.com> at Sun Nov  3 21:17:20 2002 (2013 days); it has been requested 61 times, last by troubled, 26s ago.
16:59.11TimRiker~forget b
16:59.11infobotTimRiker: i forgot b
16:59.17troubledhehe, read this carefully:  can search on factoid strings shorter than 2 chars
16:59.29troubledwhats the only valid non 0 length less than 2?
16:59.30TimRikersure enough.
16:59.50TimRiker.5 ?
16:59.53troubledheh
17:00.02troubledprobably could reword that one
17:00.11troubledthat or make it <=2
17:00.11TimRikerperhaps. :)
17:00.26troubleddamn man, how the bot get this bad hehe
17:00.29TimRikerinfobot: chanset #infobot -minVolunteerLength
17:00.29infobotUnsetting minVolunteerLength for #infobot; was '3'.
17:00.35TimRikerbzflag?
17:00.38troubledthere is sooo many places to start heh
17:01.24troubledso 3 is a good min for you? it misses all the smilies so I like it
17:01.28TimRiker"o" can't reload? need "M" ?
17:01.37troubledfile that under "oops" :)
17:02.02troubledshouldnt need +M to reload. where you getting that error?
17:02.20TimRikeras a default? we need to decide what default we will use to get the "set addressing REQUIRED"
17:03.11troubledwell, setting min and max to something higher than the max msg like 1000, should force the check to always fail
17:03.40TimRikerthere should be a way to get help on all the channel flags. just a local script would work, but over irc would be better. not thought that through yet.
17:03.49TimRikerI always end up grepping the source.
17:03.56troubledi know the feelin :)
17:04.12troubledctags helps for some stuff though
17:04.33troubledthink its like ctrl-t in vim to get em to jump or something
17:04.42TimRikerthe bot can grep the source too, if we add a command for it.
17:04.52troubledthats a security nightmare imho
17:05.44TimRiker"chansethelp" or some silly thing. just "grep -Irs src | grep -v .svn" after stripping [^a-zA-Z_] or some such.
17:06.05TimRikerprobably 0-9 in there too.
17:06.06troubledoh, you mean a command users can run to get help?
17:06.17troubledsomething that just runs grep or whatever?
17:06.34TimRikerif they like. mostly for admin use, but might as well let the users do it too.
17:06.55TimRikerit's open source after all.
17:07.01troubledwell that or cleanup and finish all the loose howtos and faq docs
17:07.15TimRiker~srcgrep minVolunteer
17:07.43TimRikeronly allow one line of bot reply.
17:07.49troubledlets save new features for later :)
17:07.52TimRiker:)
17:08.13troubleddont get me wrong, if we need something new like a var or conf setting to fix something, by all means heh
17:08.16TimRikerso, min/max could be unset in _default, and required for volunteer.
17:08.34TimRikerthen the example has then set, perhaps to silly values, in #infobot etc.
17:08.41TimRikerthoughts?
17:09.13troubledwell, don has expressed that he expects the bot to always behave the same as far as backward compatibility goes or something like that
17:09.34troubledso adding and changing stuff is fine, but the defaults should match something not so chatty i guess
17:09.53TimRikerquery should already check some max non-volunteer length, so max is not required. but min could be unset to disable, and set to enable volunteer.
17:09.53troubledie: _default of 1000/1000 (with explanation to user somewhere on why)
17:10.25troubledthere is no maxVolunteer check in the code that works for that hungry mode remember
17:10.32TimRikerunset would mean migration is sane.
17:11.03troubledi was thinking of using that getChanConfOrDefault() though
17:11.06TimRikerin your 1000/1000 proposal, what happens if they are not set at all?
17:11.21troubled^^^ that would set them to 1000 if they arent set
17:11.43troubledthen we can update the default conf, and still allow us older bots to run and get the settings on next file db saves
17:11.57TimRikerin each and every channel? oh. just return them if not set. /me ponders
17:12.27troubledno, i mean there is a sub that actually does that already. get the conf value, and if its not set, set to the second param as the default
17:12.44TimRikernods. that's what I'm pondering. :)
17:13.09troubledIRC/Irc.pl:        my $j = &getChanConfDefault( 'sendPublicLimitBytes', 1000, $chan );
17:13.15TimRikerif someone sets min to 5, then max could remain 1000. that's bad.
17:13.29troubledsome parts of the bot use that, some dont, thats another issue that needs to be addressed :/
17:13.38TimRikernods
17:13.58troublednah, max of 1000 is fine, just means it would always check the db regardless of size
17:14.23troubledthat or we could just assume -1 to be a sentinel value with special meaning of disabled or N/A or something
17:15.12TimRikerwe don't really have a max query length now that I can see.
17:15.16troubledso something like: &getChanConfDefault( 'minVolunteerLength', 3, $chan ); and -1 for max
17:15.22TimRikerjust whatever the server will send us.
17:15.33troubledthats actually another design decision, what max do we really want long term?
17:16.03troubledive seen yours or don's db with 1000char entries that appear to be manual imports. Do we ever want to allow pasting that stuff split up to a channel?
17:17.33TimRikerif length $query > &getChanConfDefault( 'minVolunteerLength', 1000, $chan )   would do it atm, no?
17:17.45TimRikeror in this case, return if <
17:18.17troubledthink so
17:18.27troubledwait, < not >
17:18.33TimRikerwould be nice to have a $maxquery someplace and replace that with &getChanConfDefault( 'minVolunteerLength', $maxquery, $chan );
17:18.44troubledyou want to return is the lnegth is less then the min, or greater then the max
17:18.49TimRikerright.
17:19.02troubledi think a maxquery is in there somewhere isnt there?
17:19.19troubledtechnically its limit to 1 msg in irc though which is 450-500 typically
17:19.44TimRikerthere's a maxKeySize, maxDataSize ...
17:19.57troubledim probably thinking datasize
17:20.34TimRikermaxlinelen
17:21.01troubledi think that one must be left over from the 1000char+ factoid tests
17:21.24troubledcause i cant see how its even logically possible to have that when irc is 1 msg per factoid atm
17:21.53troubledbut i was going to ask if that was ever desired though. If so, could be handy to keep that around
17:22.16TimRikerclearly some replies are multi-line.
17:22.43troubledgenerally module forkers, but those arent governed by that are they?
17:22.47TimRikerit would be nice to have a unprivileged user limit, and then an op limit perhaps. don't care much though.
17:23.22troubledbeen meaning to ask about that. policy needs to be made with respect to chan ops
17:23.27TimRikernope, but they might end up referencing any variables we set up, etc.
17:23.35troubledshould they get implied bot access? if so, what kind? how much?
17:24.07TimRikernot much. :) topic editting and other things that chan ops would have.
17:24.21TimRikerif they need more, they should get registered with the bot.
17:24.25troubledthe way i look at that is that most setups are probably modeled around a hosting setup. ie: you host a bot for a channel. thus chanops should have access for their chan only, but you being the owner have full access to all
17:24.40TimRikerthough it might be nice to "trust" chanserv permissions as an option.
17:24.54TimRikernods
17:25.28TimRikerie: if you're an op in #debian, you can edit #debian channel settings.  that would be nice.
17:25.39troubledthat would be awesome, but keep in mind that 1) chanserv is thiea here atm and moving to atheme soon and 2) not every network has a chanserv and 3) not every chanserv is ame output
17:25.41TimRikernot _default settings, just #debian.
17:26.07TimRikerwell, I mean trusting chan +o however that happens.
17:26.21troubledwell, one interesting bot ive seen is in botpark. Anna^
17:26.36troubledwhat it does when you op is add you to its access list and inform you
17:26.52troubledits not perfect or very useful to me, but the idea has some merit I guess
17:26.55TimRikerpermanently?
17:27.05troubledwell, thats a bug I think, not atm it doesnt
17:27.18troubledevery time I op it adds me. but that could be his design
17:27.33TimRikerwould be nice to cache chan +o +v on the userlist and use that in a few places I suppose.
17:27.35troubleddont think it forgets me til I part chan or quit irc though
17:28.06TimRikerif hasChanFlag('o') etc..
17:28.20troubledwe need to approach this patches in a more organised fashion. i sorta have an idea though if you give a sec
17:28.43TimRikerlistens
17:28.52troubledwe want to head towards a rewrite (seems to be POE atm, but who knows right). What about replacing all the guts one part at a time?
17:28.57troubledcontinues....
17:29.49troubledso lets say we start on conf chan settings. First step, pick a var, grep it in all source and erase it, renaming it to something "proper". Ie: has a naming convertion, is strictly followed etc. then...
17:30.32troubledwe do the same to the next var and repeat until the one file is done. then move on to something like subs and make sure say all subs are of that getChanConfDefault type (goal being consistency)
17:31.01TimRikersounds painful for current bot owners.
17:31.05troubledeventually we would end up with a new vars. the next step after than could be to start enabled strict across the board for example
17:31.32troubledwell, conf vars might not be a great start. maybe subs or something thats abstracted away from the owners for now
17:31.41troubledbut for vars im thinking, no more case sensitive!
17:31.48TimRikerI've enabled strict in quite a few places. I tend to do that when I'm making larger changes to a file.
17:32.05troubleddoesnt mean we cant go hungarian or something in the src, but if we at least allow some flexibiblity, it could help
17:32.39troubleddo you see a reason against case insensitive for config options where they make sense?
17:33.00TimRikerI'd prefer just lower case them all and _require_ lower case. you want to be able to look them up in a hash without having to lc it each and every time you do a lookup.
17:33.13TimRikerperformance is the reason.
17:33.15troubledobviously, something like a password should be sensitive, but search for password/Password/PassWord should all find the same var
17:33.25TimRikerdisagree.
17:33.30TimRikerpick one and use it.
17:33.40troubledlowercase could be fine too, but that doesnt address the owner change problem
17:33.48TimRikerI can see the argument that _all_ chan vars are lower case.
17:33.58TimRikerowner change problem?
17:34.08troubledim thinking more for the users benifit. the coders are always on their own to lookup stuff
17:34.25troubledowner change = when they update and all their config are belong to us :)
17:34.28TimRikergrep -i is a marvelous thing.
17:34.54troubledim fine with grep and all that, but we really should make it earier on the bot owners that arent devs :)
17:35.01TimRikerah, if we do the lc conversion, we should have a lc on the .chan loader. NOT in every lookup call.
17:36.35troubledas for speed, i dont think it applies for conf files for us
17:37.14troubledif they are all in a cache hash that is searched with lc, whats in the file isnt going to matter much. files are only ever touched when loading or saving
17:37.29troubledits the same reason freenode uses a flat text file to save the user db's
17:37.44troubledthey just dump ram to a file fast as possible and get it over with
17:38.00TimRikerwe du multiple lookups for each line in each channel the bot is in. it _does_ matter.
17:38.06TimRikers/du/do/
17:38.08troubledsure a dbm is gonna be faster for disk searches, but that isnt our case
17:38.32troubledthats gotta go then, the settings should go to ram and stay there til i !save
17:38.50TimRikerif we lc the vars on load, and then save, the file will be lc. as in a .chan file.
17:39.01troubledideally I think we _need_ everything to get into var's and hashs in prep for POE/OOP design
17:39.15TimRikerit _is_ in a hash now.
17:39.36TimRikerbut I don't want to add a lc on every hash lookup. there are too many of them.
17:39.41troubledwhile there are a few things like my rss feeds that do file reads per invokation, I think its best to get it all into ram and periodically flush to the backend
17:39.48troubledah
17:40.01troubledthought you were saying the .chan file was read every lookup heh
17:40.08TimRikernope.
17:40.39troubledyou cant decide between speed or features can you :)
17:41.28TimRiker%chanconf is the hash
17:41.29troubledwell, lets look long term. where do we ultimately want the chan conf's to be saved? in a db? if so, the time and effort to change vars might be best spent getting a skeleton POE setup
17:41.38TimRikerI was speed _and_ features. :)
17:41.43TimRikers/was/want/
17:41.56troubledwell, features all gonna have to be forkers then :p
17:42.27TimRiker.conf should be a cached hash of an sql table. it's now a caches hash from a txt file.
17:42.46TimRikers/caches/cached/
17:42.48TimRikersheesh.
17:42.50troubledstill, even in a db, case will need to be addressed sooner or later i guess
17:43.32TimRikerdb, table, file can all handle mixed, upper, lower, case insensitive. that's not the issue.
17:43.50TimRikerthe issue is, does the mixed case make an admin's life easier or harder?
17:44.03troubledno, the issue is we have to decide which to use, mixed upper _or_ lower, not all 3 like now :)
17:44.06TimRikerif it's easier, keep it. if it's harder, throw it out. :)
17:45.03troubledthis would all be easier if all the conf var's had a registration hook ala commandstubs for registering their help and conf var's
17:45.11TimRikerseems to me that mixed case causes more headaches than it saves. I tend towards lc only. no '_', just multiplewordsstucktogether
17:45.52troubledi like that just fine too. and its easy patch to auto convert stuff from old dbs to lower case on flush once
17:46.04troubledwell, s/once/every flush/
17:46.34troubledbtw, did you ever look at immute's poe setup?
17:46.50TimRikerjust a line in the source someplace like DefineChannelVariable: minVolunteerLength - how short a line do we try to volunteer replies to? 1000 or so to never reply
17:47.09troubledi started to model some of that test branch changes using his ideas, but i dont like some of the layout. mostly he has the right idea though
17:47.14TimRikerand then a script/command to grep for those would suffice. it's not like we do that often.
17:47.38TimRikerNOT every flush. just on load.
17:47.57TimRikernot looked at immute's poe stuff, no.
17:47.58troubledyou would need it to catch any new stuff they add thats wrong
17:48.21TimRikerwe'll catch that next load anyway.
17:48.32troubledso if i start a bot and i did "chanset _default SomeThingWongHere", it would lc it either when I set it, on on flush. on set makes more sense i guess
17:48.40troubledbut then its broken until reload
17:48.51TimRikerif you want to handle op adds, better to return an error when they try to "chanset _default SomeBrokenThing"
17:48.56troubledDynaConfig.pl does realtime stuff lke that iirc
17:49.12TimRikerdon't fix it, just tell them about it and train the admins.
17:49.44troubledif we want that, we best have conf var registration then, which would be fine but need a full src grep for all conf var's that need to be reg'd and add hook on startup to call them
17:50.39TimRikerconvert the existing code, train the ops in irc, convert on load of old configs. that's all that should be needed.
17:51.04troubledwere you just thinking of an error that says "sorry, must be lowercase"?
17:51.35TimRikerif someone adds a module that checks for "SomeBrokenThing" then it's never going to be there, so as a developer, they should notice that. they will try to set it in the bot, and get the error then.
17:51.58TimRiker"sorry, all channel options need to be lower case", yes.
17:52.24troubledin that case, why not just tell them it should be lower and that you are going to convert it for them?
17:52.40TimRikercause then they can ignore the message and not get trained.
17:52.41troubledjust seems like we should take care of mundane details like that for the user
17:52.58troubledtraining eleet bot ninjas are we? :)
17:53.05TimRikerexactly. :)
17:53.22troubledso on the note of POE....
17:53.32troubledyou like it? use it much?
17:53.45TimRikerI want "oh, duh. I have to type that again. Guess I won't do that wrong again" instead of "I can ignore that silly comment"
17:53.52troubledanything in particular that you know we cant use in it?
17:54.36TimRikernot used it much, no. having a bot that can be on aim, gtalk, irc, web, sms, etc would be a good thing. if POE gets us closer, and works, then I'm for it.
17:55.07troubledwell, ive made a bot from the same code and i know the irc server and the poe client work fine for basic irc
17:55.11TimRikeras far as I'm concerned, if the perl modules are in debian stable, we can use them. :)
17:55.27troubledhowever, i did have one person complain that POE's flood controls arent always perfects and sometimes flood offline
17:55.46troubledbut POE you can inhrit, and override that with something custom if it sucks that bad. just more work
17:55.48TimRikerthat's no biggie. irc flood controls aren't perfect either.
17:56.16troubledwell, for poe i was really hoping to just kinda rip out the core and plop it into POE with as little edits as possible at first
17:56.19TimRikerI want forkers to hand replies back to the parent. NOT just write to the same file handle.
17:56.50troubledbut another real big goal I want to aim for is making an actual OOP design for calling and setting var's, flushing files etc
17:56.53TimRikerso we get one process knowing about stats and flood requirements.
17:57.05troubledya, POE should solve the inter sub communication problem 100%
17:57.16troubledevents are the way to go imho
17:57.24TimRikerugh. not a fan in general. show me specific advantages and we'll talk.
17:57.49troubledcleaner code come sto mind :)
17:58.02TimRikerfor a project like this all OOP adds is more levels that you have to dig through to see what's really happening. we have too many of those as it is.
17:58.09troublednot to say you have to use events for everything. modules can stay as called sub's i guess
17:58.26TimRikerwe need to strip the layers down, not add more ways to find layers so we think our code looks clean.
17:58.30troubledbut I think you run the risk of a sub hogging cpu time though
17:59.22TimRikerthat's why we have forkers.
17:59.31TimRikerif it can take a while, forke it.
17:59.35TimRikers/forke/fork/
17:59.59troubledtake a look at this irc client: http://poe.perl.org/poedown/oscon2004/chat-client.perl
18:00.22troubledbut the whole idea behind the even system is that you dont need forkers anymore, its builtin i guess
18:00.57TimRikerclearly "$heap->{vt}->print($heap->{window_id}, $input);}" is cleaner.
18:01.10troubledinstead of a fork, you just call the event and when the sub finishes, it posts its value back in the form of another event that the core event handler would then redirect. seems very smooth overall for my db setup
18:01.35troubledyes, that is clean :)
18:01.35TimRikerand it the sub never finishes?
18:01.40TimRikeror takes 20 minutes?
18:01.49troubledthen it takes 20 minutes
18:01.57TimRikerand we drop off irc?
18:02.00troubledpoe makes it all multitask behind the scene for us
18:02.17troubledthen an event fires with the irc code and we handle it specifically
18:02.34troubledhttp://poe.perl.org/?POE_Cookbook
18:02.45troubledhttp://poe.perl.org/?POE_Cookbook/IRC_Logfile_Tailing
18:03.04troubledspecifically that one i know works good. it shows the irc raw's and how to hendle them specifically and or generically
18:03.29troubledthe yield and delay subs are what allow backgrounding iirc
18:03.45TimRikereek.
18:03.48troubledone waits for a return (yield) and another just sends the event and continues
18:04.24TimRikerso we are running on a true multitasking os. we now choose a time sliced implementation so we can avoid real multitasking that the os gives us? that's an improvement?
18:04.32troubledwhen i first looked at poe and played, ill admit it was scary, but its actually very well done framework imho
18:05.22troubledkeep in mind its all OOP. we would have full control over what gets run and how
18:05.34dondelelcaropoe is definetly the way to go
18:05.48troubledheck, nothing stoipping us from just running the whole bot as is right now in a pipe in poe
18:05.50dondelelcarohas written about 4 other bots using it
18:06.12TimRikerdondelelcaro: so convert us already. :)
18:06.32troubledwell, i was playing around, but that was hardly a poe rewrite start
18:06.51troubledand immute had a good start but thats nowhere near backward compatilble with the direction he wants to go i think
18:07.00dondelelcaroyeah
18:07.07TimRikermy point is the simple example I've seen don't do any blocking external io in any of the methods. they DO do that in POE itself. ie: connecting to multiple chat networks and not hanging on one.
18:07.11dondelelcaroI personally haven't had time to even start into it
18:07.30dondelelcaroTimRiker: you shouldn't ever do blocking external IO; that's just bad.
18:07.40TimRiker_we_ do external blocking io in forkers. I don't see a POE solution to change that bit.
18:08.17dondelelcaroso you fork, and use one of the IPC modules to deal with the forkers
18:08.28TimRikerdondelelcaro: if your request needs to hit an external we page to get data, that's blocking. no matter how you slice it.
18:08.38dondelelcaroTimRiker: uh... no, it doesn't.
18:08.48troubledwith POE its abstracted from us though
18:08.54dondelelcaroTimRiker: there are modules already written for poe which handle getting select requests without blocking.
18:09.01TimRikeryou need to either fork, or have the base c level select() listen on all your external connections.
18:09.11dondelelcaroTimRiker: which is something that POE already does
18:09.15TimRikerdns, local files, tcp connections, udp requests, etc.
18:09.21dondelelcaroyes; poe does all of those
18:09.22troubledall taken care of
18:09.24TimRikeras does NET::IRC
18:09.31dondelelcarowe won't use Net::IRC anymore
18:09.39dondelelcarowe'll use PoCo::IRC
18:09.51troubledbasically if we want to be "professional" about it, we would extend the bot with another custom POE module and then just use it
18:09.52TimRikerthe question is not if it does it, it is: is it clean, easy and reliable.
18:10.05dondelelcaroTimRiker: it is clean, it's easy, and I've not had a problem with it yet
18:10.26TimRikerwith Net::IRC, it's reliable, but not clean or easy. and not integrated with our lame shm implementation.
18:10.30dondelelcarothe horde of pastebots I wrote for the debates is a prime example of running a slew of bots in a single thread without any blocking
18:10.39dondelelcaroit's not even reliable, unfortunatly.
18:11.00TimRikerok, so lets say someone asking for factoid "foo" and you look it up in mysql, and mysql blocks. how does poe handle that?
18:11.01troubledthat and you solve a big problem or two right off the bat: readline capable console, working dcc's and removal of a few files of useless code that isnt needed anymore after those 2 things get fixed
18:11.16troubledTimRiker: poe just keeps going until the even comes back to you
18:11.31TimRikertroubled: what event?
18:11.32troubledyou act on the event, not the lack of one :)
18:11.46troubledthe modules will eventually time out, all that stuff is generally setable in modules
18:11.52TimRikerwhat event?
18:12.03troubledbasically POE would nudge us into written proper code and perl modules in an OOP fashion
18:12.05dondelelcaroTimRiker: when the query finishes executing
18:12.15TimRikerdondelelcaro: that's NOT an event.
18:12.26dondelelcaroTimRiker: uh... sure it is
18:12.29troubledTimRiker: with the db stuff specifcally, I believe you can set timeouts which would then gerate an event that you can handle
18:12.29TimRikerthat's a ->query(). NOT an event.
18:12.42troubledbut poe makes it an even :)
18:12.45troubledOOP!
18:12.53TimRikeror does POE replace the DBI interface?
18:13.22troubledwell, they abstract it
18:13.25dondelelcaroTimRiker: it wraps it, yes.
18:13.27troubledsec, ill fecth the link
18:13.54troubledhttp://poe.perl.org/?POE_Cookbook/DBI_Helper_Processes
18:14.01TimRikerok, on to the next... I update an RSS page of the net. they wrap that too? I hit a udp port on the bzflag server to get player info, that too?
18:14.03dondelelcarothere's DBIAgent, LaDBI, and a few others
18:14.25troubledthats actually the code i based my echelon ssh login off of. and believe me, its smooth considering my irssi blocks hard on netlag and poe doesnt
18:14.30TimRikerI read a grep a local file for larts and the filesystem is blocking for a moment, that too?
18:14.57TimRikers/read a grep/grep/
18:15.03dondelelcaroTimRiker: all of that can be dealt with, and it's not like we can't just write a module which forks off the plugins when they're being executed and handles shooting the events back to poe via IPC
18:15.32dondelelcaro(if we want to maintain backwards compliance without large rewrites)
18:15.56troubledTimRiker: spent some time tonight and just glimpse over all the stuff they show on this page: http://poe.perl.org/?POE_Cookbook
18:16.00TimRikerso basically, every perl method you know of that can block, needs to get killed and replaced with some POE equivelent that the average perl coder is not familiar with, or the hole thing falls apart?
18:16.03TimRikersounds like a plan.
18:16.13troubledthere is examples of every aspect of the bot needed right there if you can copy/paste well :)
18:16.23dondelelcaroTimRiker: uh... if you're going to write a bot using the framework, you have to know the framework.
18:16.51dondelelcaroTimRiker: it's far better than now, were every time you have a routine which blocks, you either have to be in your own fork, or you have to deal with not blocking yourself.
18:17.08TimRikerseems like a lot of work to re-invent multitasking. it's back to the window 95 for you!
18:17.23dondelelcaroTimRiker: uh... the entire point is that we're not re-inventing it
18:17.23troubledor the routine now where I want to fix 1 bug, and have to rewrite have a bot cause each bug is dependant on another one :)
18:17.33troubledhalf*
18:17.39dondelelcaroTimRiker: what we have now in infobot *IS* us re-inventing it.
18:18.23dondelelcaroin any event, it'll become clear once someone actually has time to start rewriting it
18:18.31dondelelcaro(which won't be me for a considerable length of time)
18:18.31TimRikertrue that. I can't argue that today is not very clean. but if a large rewrite gets us to the point where we can still get blocked by that one io event, we've not improved much.
18:18.48troubledwell, i could lay out a somewhat compatible "get on irc and sit there stupid" bot as a start anytime really
18:19.03dondelelcaroTimRiker: any framework to avoid io problems *always* requires that programmers be aware of io blocking
18:19.38dondelelcarothere's no magic bullet that avoids any blocking on IO if you're stupid or ignorant when you're coding.
18:19.50troubledTimRiker: you ever seen turbo pascals Turbo Vision framework? or Turbo Vision for Borland Turbo C?
18:19.50TimRikerdondelelcaro: some programmer, yes. but the advantage of the current, somewhat broken, forker implementation is that only the core developers need to be aware of it. the module writer can block all she wants.
18:20.04dondelelcaroTimRiker: 11:15:03 <+dondelelcaro> TimRiker: all of that can be dealt with, and it's not like we can't just write a module which forks off the plugins when they're  being executed and handles shooting the events back to poe via IP
18:20.22TimRikertroubled: yes, was also a polling tasking environment from what I recall.
18:20.27troubledTimRiker: we can keep that design technically, but POE tends to make it better to just redo it as a poe module
18:20.45dondelelcaroTimRiker: I'm not suggesting that we force the plugin writers to be aware of POE; I'm only talking about how the internal modules are dealt with.
18:21.04TimRikerif it's just the same, I don't understand why that's better. and enough so to justify doing it.
18:21.28dondelelcaroTimRiker: because it means that the internal bits of the bot won't block on itself
18:21.50TimRikernods
18:21.52dondelelcaroright now, you have to be absolutely certain that you don't block anywhere in the main thread
18:23.14TimRikerthe only large blocking bits I know of in the internal bits is the sql issue. hence I run sqlite which minimizes that as compared to mysql/pgsql.
18:23.14TimRikernot removes mind you, just reduces.
18:23.14dondelelcaroyes...
18:23.32troubledso right off the bat, just going poe = dcc works again, no more shm that we need to deal with and a fully functional console _with_ tcp access as well. and thats just moving the core into poe and keeping everything blocking to the extent it does now. that sounds like a pretty good start :)
18:23.50dondelelcaroyeah
18:23.59dondelelcaroplus, PoCo::IRC is actually maintained
18:24.07troubledtcp could be a telnet admin login, or a web server or heck, an irc server :)
18:24.15TimRikerso I'm saying I _do_ see a clear advantage to putting the sql pipe into the central select(). Net::IRC makes that very difficult. not impossible, but difficult and not supported.
18:24.16troubledya, BingoS i think does it doesnt he?
18:24.26troubledand he hangs on freenode, cant go wrong :)
18:24.48troubledTimRiker: also, POE can mean windows/mac
18:25.04TimRikerNet::IRC works on windows/mac
18:25.07troubledwindows support has limits of course, so it would be restrictive, but probably be ok
18:25.43troubledtechnically you can extend poe to use whatever you want. but that requires Turbo Vision like skills with POE to know how to make a "proper" modules
18:26.28troubledthats what im trying to say, over time if you slowly convert stuff to poe modules, eventually you will have a beatuful code base that you could even probably pop into cpan for some modules
18:26.28TimRikertroubled: with Net::IRC tcp can be anything too. telnet/ssh/etc can all be based into the central loop. _we_ don't do that, but it is supported. well, at least as far as Net::IRC is supported.
18:27.19troubledTimRiker: tbh, im mostly interested in a OOP design, POE happens to be a framework already make that is more up to standard than i or any of us could ever accomplish in a sane amount of time
18:27.44troubledits spit polished :)
18:28.25troubledsure we could just split the bot into perl modules and OOP it ourself and probably make it faster and cleaner than it is now, but thats a lot of work compared to poe I think
18:28.49TimRikerI'm not a fan of OPP just to add a buzzword to the project description. in my experience OOP == more folks working on the code that have no understanding of performance and therefore a slow griding loop of death till only a super computer can possibly get simple tasks done.
18:28.55troubledi just think poe lends it self nicely to gradual transition, which is probably best for all of us
18:28.57TimRikeroh, but the code _looks_ clean.
18:29.34troubledfor some reason, I doubt any of us will ever have a db so stupidly huge that the rest of the machine lags from POE alone :)
18:29.57troubledlets face it, this bot is really small resource wise
18:30.29TimRikerit's not the db that's the issue. it's noticing that your one small line of code added actually does a million db lookups for each line of irc traffic.
18:30.51troubledhow is that poe's fault?
18:31.09TimRikeryou still have that issue with a non-OOP design, but you're more likely to be aware of it then.
18:31.46troubledbasically an irc msg will generate the event. in that even we basically copy/paste the old logic to call the other subs as a temp solution. it has blocking issues where we dont use forkers, but its still temp until stuff is converted to handle poe properly
18:31.54TimRikerthe more you hide in alleged 'clean' code, the more you hide the performance impact.
18:32.12TimRikerand memory impact, and indirection overhead, etc... all performance related.
18:32.17troubledim just sayin, poe isnt going to magically force us to query every line in the db. we would have to tell it to do that
18:32.33TimRikerthere are good reasons why kernels are not OOP for example.
18:32.42troubledtheres no way performance could hurt these machines enough for anyone to notice
18:32.55troubledpolitical mostly :)
18:32.55TimRikernow, I'm not saying this means no OOP here. I'm just expressing my concerns with it.
18:33.14troubledthat and the kernel is a bad example since they need speed at that level
18:33.21TimRiker"<troubled> theres no way performance could hurt these machines enough for anyone to notice"
18:33.35troubledi was talking about any one of our infobots
18:33.52TimRikercorrect me if I'm wrong, but aren't you the same guy that was concerned with 2 db lookups instead of one a while back?
18:34.19TimRikerso didn't you "notice" and wasn't that a performance issue?
18:34.30troubledwell ya, im not just gonna double lookups for nothing heh
18:34.38TimRikerand if so does it not make your statement completely false?
18:34.59troubledthere is a difference between a worth wile sacrifice and an outright dumb design flaw :)
18:35.57TimRikerand if OOP leads you to clean lookig code that just happens to double your db lookups, is it still worth it?
18:36.05TimRikers/lookig/looking/
18:36.36troubledit wont though
18:36.55troubledjust cause its POE doesnt mean the db lookups will change whatsoever
18:37.42troubledhttp://poe.perl.org/?POE_Cookbook/IRC_Bots
18:37.46troubledrun that ^^^
18:38.25troubledwell, later perhaps, but try it. get yourself poe installed, hit up that site for some examples and see how they run
18:39.23troubledthe tutorials on that site are great showing what POE vs. non POE look like. but you would be hard pressed to get a fully irc client in that many lines of code without constantly redoing everything that PoCi does for ya
18:40.58TimRikertroubled: yes, I've looked through the simple non-error handling non-blocking method examples. They are all well a good. nice and simple. but the Net::IRC examples also meet that criteria. that's NOT where my questions lie.
18:41.45troubledthe poe modules typically do have error handling all takin care of remember
18:41.59TimRikertroubled: notice I did NOT say that POE could impact the db lookups, so nice unrelated comment you made.
18:42.18troubledwell, by taken care of I mean they generate an event for every possibility the underlying api supports. its up to us to catch and handle the events of course
18:42.37TimRikerthat's no different that Net::IRC claims.
18:42.39troubledsorry, but you did mention double db lookups :)
18:43.02TimRikeryes, but NOT because of POE, which is what your unrelated comment was about.
18:43.35troubledwell, poe is really only a tiny part of the whole PoCi system
18:44.19troubledits base functionality isnt anything more than some well planned and written perl modules. no reason we couldnt just do similar, OOP or not
18:44.57TimRikermy statement was about OOP. if you get 'clean' code, you normally mean that you hide actual code with a slick layer on top. and my comment is that that often means the code tends to get less efficient because the interface does not look like it will have performance issues.
18:45.08troubledwheither I call a sub called setAge() or getAge() doesnt mean much compared to $obj->getAge()
18:45.54troubledI guess I just like oop in that sense since it brings everything into a nice easy to find organised place
18:46.03TimRikerI did not say that performance is the only critera. nor did I say that POE requires a complete OOP design, nor did I say that it _requires_ a less efficient design. all I said is that OOP tends to be less efficient.
18:46.08dondelelcaroOOP isn't magical, but it sure beats the crap that we have now.
18:46.16troubledaye
18:46.36troubledtbh, I just want uniformity. oop seems like a good way to go to get it
18:46.37dondelelcarosensible use of OOP, coupled with sensible use of other methods is the way to go
18:46.50dondelelcarouse OOP where it makes sense; don't where it doesn't.
18:46.55troubledexactly
18:47.10troubledwhich is why i was thinking if its gradual that the forkers dont matter at first
18:47.21troubleduse POE? so what, fork em still :)
18:48.53troubledif I did start a rewrite, my intention was to have a core oop design with all the varibles and methods to get and set them using the autoload way
18:49.25troubledsaves lots of code for writing typical wastful and boring getProp/setProp() subs that take up 100 pages
18:50.22TimRikeras to poe specifically... cooperative multitasking is generally less reliable and less efficient that preemptive multitasking. that does _not_ mean it is to be avoided at all costs. it's just that from my inexperienced perspective with POE it seems like it leads you down the patch that all is well, when with any cooperative multitasking implementation, all it takes is one uncooperative member, and you have a big pile of crap.
18:50.43troubledhttp://pastebin.ca/1012977
18:51.05TimRikercourse it might be a better pile of crap than some other perhaps non-oop pile of crap. :)
18:51.35troubledtheres what I use now and is probably a design i would start with. Yes its redundant in that pm's out there already do better, but its small size, easy and since its OOP, it could merge into POE easily
18:52.40troubledi know what you mean, but I think that good use of poe could avoid problems. and forking is always an option for stubborn code or old buggy code no?
18:53.04TimRiker#
18:53.04TimRiker<PROTECTED>
18:53.04TimRiker#
18:53.04TimRiker<PROTECTED>
18:53.20TimRikerstill compltely blocking. so what was the point in that?
18:53.39troubledits clean and organized?
18:54.03troubledtrust me, the whole logging setup i have now is slow and blocks. thats just my inexperience and lazyiness:)
18:54.07TimRikerand does not handle all the differences that our current pile handles?
18:54.29TimRikerif we throw away all the exceptions we handle, we'd be clean too.
18:54.32troubledthats just a sample of how I want to lay it out, not the actual methods and var's im gonna use :)
18:55.03troubledi know my pm's are far from handling everything, but i dont care since im the only one using my setup so its diff then infobot
18:55.04TimRikermy point is that once you add in all the odd cases we have now, it will end up looking mostly like what we have.
18:55.28TimRiker"here's a simple example", see? it looks simple.     well duh.
18:55.35troubledpossibly, cept alot of cruft and bugs would be fixed as it gets converted
18:55.56troubledif you built it in layers slowly, I think each subsystem would get a nice overhaul
18:56.25TimRikerand a lot more introduced too.
18:56.34troubledless I would think
18:56.51troubledalot of code is just copy/paste with adjustments for path changes and such
18:57.18troubledbut by layering it system by system, you can probably do major overhauls on each peice as you set it into its final resting place in the new code
18:57.37TimRikerheh. start digging through the current code. it's got to be complex over time as more features and requirements were thrown in. if you don't know all those at the start (and nobody I know does) then a new design will end up with the same issues.
18:58.11troubledthats sorta while I would like to start the rewrite. I cant work on a bot I dont know, and I cant know a bot better than rewritting it myself
18:58.27TimRikerthe way to clean code is to track requirements and implement to match. if you don't know the requirements, you can't cleanly implement them. oop or no oop.
18:58.34troubledso the problem then shifts to the other devs (that'd be you guys) to keep up and follow along
18:59.15TimRikerwell, don't let me stop you, or anyone else for that matter. :)
19:00.21TimRikerlearning is justification enough. but... I like my bot working. so moving to something that works less, is going to be tough to sell. not impossible, just tough.
19:00.32troubledwell, i can make a branch and start playing around, and i was starting, but I ran into the global var scope problem. Basically I want to address that with a oop setup using autoload for var's and then plug stuff back into it from there, grepping and sed'ing subs to reflect changes as i import stuff
19:00.59troubledwell lets face it, you wouldnt want to run any test branch I or anyone makes until its ready for public testing at least
19:01.37troubledone requirement to add I guess would be that all you need to do is svn switch to the branch and pooof, the bot just works like before (as a start)
19:02.08troubledand immediately after you would start changing stuff that might alter syntax or remove old features, add new commands etc.
19:02.20troubledat least that seems like a good approach to me
19:02.48TimRikerwell, I think that requirement could be improved on. :) there are some "features" that could hamper a new design. the issue is convincing someone that likes the current implementation that the new one is worth moving to.
19:03.37TimRikerso I think the two need to evolve towards each other. the current getting cleaned, and the new adding features.
19:03.55TimRikerat some point the move becomes easy enough that it's worth it.
19:04.08troubledsorta what i started working towards the other week
19:04.51troubledbut rather than fix the stupid broken globals/locals, i was just gonna rip the damn things out and put them in a central file and patch it all to work again. lather rinse repeat with another subsystem
19:05.25troubledif all went well, that change alone could be done in a day prolly
19:07.05troubledreally its just a matter of say an autoload, then grep for every instance of each var you move to the new file across the whole code base and replace with the new sub/oop syntax. not really much than could go wrong there that I can see
19:07.32troubledi mean either you typed it correct and the bot runs, or you had a typo and it crashes or doesnt run right
19:11.03troubledbtw, you know the CIA bot here tight? Thats just a python script, with like 50 or 60 nicks, dosens or servers and hundreds of channels, all on 1 machine and, for the most part, 1 process
19:11.52troubledruns like clockwork, and likely to be bigger load than anything we will ever see with the bot. and I cant see python being much diff than perl for our usages
19:12.26troubledit uses xmlrpc to communicate between its children though where as poe im not sure what they use internally
19:25.21troubledTimRiker: anyways, good chatting as always. Very exhausting to discuss all this. Ill leave you alone for the rest of the day to go about your business and read on POE or something. As for that patch, ill probably hold off on it in the mean time since you found the workaround for hungry mode. We did cover alot of bases though so I if I do play around, at least I have an idea of where we all want to accomplish with a cleanup. catch yo
19:26.19TimRikerwell, I dunno about clockwork. it's been down often. but dunno if that's more or less than my bot.
19:26.42TimRikeris chatting in other work channels, on the phone, coding, etc.
19:30.16troubledno worries, im gabbin in blender with the guys anyways. bots been yappy since hungry mode in there :)
19:44.16*** join/#infobot Dotted (i=Dotted@bureaucrat.wowwiki.dk)
20:12.16troubledTimRiker: http://pastebin.ca/1013051
20:12.32troubledis that at least agreeable to commit?
20:13.17troubledalso, someone already requested if non addressed wouldnt work for "thanks"
20:13.42TimRiker3/1000 are NOT sane defaults.
20:14.07TimRikerdefault should be not chatty.
20:14.10troubledwell, i tested with 3/20 and it worked
20:14.32troubledbut if you get me the var name for max key size i can fill that into the max
20:15.17TimRikers/^hey([, ]+)where/where/i;   should likely be s/^(hey[, ]+)where/where/i;   no?
20:15.37TimRikeroh, s/^(hey[, ]+)?where/where/i;   even
20:15.41troubledheh
20:15.52troubledsec, ill check
20:16.46TimRikerno, I mean min should default to > max so it does not chat. chanset needed to get chatty.
20:17.54TimRikerprobably want a separate s/^(hey[, ]+)// actually.
20:18.01troubledahh
20:18.19TimRikerbefore the what/where/who stuff.
20:19.09troubleds/^(hey[, ]+)?where/where/i; didnt work when I tried it as: s/^(hey[, ]+)?where can i (find|get|download) //i;
20:19.50troubledbasically i was lopping off all the rhs stuff for questions so only the factoid key itself is left to pass on to doQuestion()
20:19.50TimRikerthe rest looks good. just need a large min by default.
20:20.18troubledwait a sec :)
20:20.27troubledwe didnt remove addressed though
20:20.48troubledso it shouldnt matter unless its running hungry like the test bot
20:21.00troubledwell, s/hungry/nonaddressed/
20:21.32troubledand for the life of me, I cant tell why that regex doesnt work. same with a few other ones :/
20:22.09TimRikermy infobot is set to: set addressing          REQUIRE
20:22.24TimRikerbut with minVL = 3 it answers.
20:22.34TimRikerso your fix breaks it. dunno why.
20:22.43troubleddid you put that last patch on?
20:22.47TimRikerno
20:22.54troubledhttp://pastebin.ca/1013051
20:23.04troubledits diff than the last one by a bit
20:23.21TimRiker&::getChanConfDefault( 'minVolunteerLength', 3, $chan ) );  is still broken, no?
20:23.36troublednope, that should work. works on my test server
20:23.44TimRikeras is +    minVolunteerLength 3
20:24.08TimRikerwhat did you fix? cause setting min=3 makes infobot chatty.
20:24.35troubledjust changed the default and added max as well
20:24.47TimRikerI don't see anything in the patch that would stop infobot from being chatty.
20:25.08troubledsec, i might need to add an $addressed check for that case still
20:25.10TimRikerand it's no chatty now.
20:25.18TimRikers/no/not/
20:25.27TimRikerwell, at least for factoid searches it's not.
20:26.14TimRikerso a default of 3 is a Bad Thing. Please change it before commiting.
20:26.21TimRikerin both places of course.
20:27.20troubledisnt 3 the min that we agreed on though? assuming it only applies when in non addressed mode?
20:27.56troubledsec, let me pop the bot into non addressed and test that patch first
20:29.35troubledya, i need an if to check for addressed mode or not before accepting "query?"
20:33.36troubledis it just me, or should the whole sub be skipped right at the start if its not addressed and we are in addressed mode?
20:36.14troubledlike "return if ( not $addressed and &::getChanConf( 'addressModeVar' ) =~ m/addressed/i);"
20:50.16TimRikerif we're not making any other changes, then a default of 50 is saner than 3. a non-default value of 3 makes perfect sense. it's having it the default that I object to.
20:53.18troubledxxxVolunteerLength is only ever active in non addressed mode
21:03.24dondelelcaropersonally, the default for it should be off, not on.
21:03.41dondelelcarosince it's almost always anoying
21:04.33troubleddondelelcaro: testing now, but the idea is the min and max volunteer settings wouldnt even do anything unless you are in addressing mode = optional
21:04.55troubledso if you stay "ADDRESSED", you wont know the bot changed. (least that is the goal)
21:05.36troubledi could probably save a shit load of if/then/else checks as well by breaking out of the sub asap for situations that dont need further parsing. checking into now
21:08.03troubled<PROTECTED>
21:08.06troubled<PROTECTED>
21:08.11TimRikerI'd prefer just setting min to something large by default. then folks can drop it if desired and turn on chatty.
21:08.28troubledeg ^^^, lots of that stuff doesnt even need to be checked if we are in addressed mode and not being addressed
21:09.01troubledTimRiker: but then you set nonaddressed mode broken by default :/
21:09.18troubledright now im going through and adding something like this:
21:09.29troubledreturn if ( not $addressed and $learn =~ m/addressed/i );
21:09.33TimRikeragreed. but "unless($addressed)" is not exactly compute intensive is it? cleaner flow would be a Good Thing though.
21:09.47troubledwell, gotta rename that var still, but thats the idea. right on line 100 of Core.pl
21:10.08TimRikertroubled: no... you get nonaddressed _disabled_ by default. which is _what we want_
21:10.36troublednon addressed? the default is already addressed isnt it?
21:11.21TimRikeryes. but with a minVL of 50 adding to that default. you are removing that setting, and hence enabling chatty mode.
21:11.42troubledset addressing          REQUIRE
21:11.45TimRikermy infobot is running in addressed only. and you saw that changing minVL made it chatty.
21:12.00TimRikerexactly. that's exactly how infobot is set.
21:12.18troubledyes thats fine then, cause im going to make the check for REQUIRE, and if its REQUIRE and !$addressed, I exit the sub asap
21:12.23TimRikerinfobot: chanset #infobot minVolunteerLength 1
21:12.23infobotSetting minVolunteerLength for #infobot to '1'.
21:12.27TimRikerbzflag?
21:12.27infobotbzflag is a 3D internet multiplayer multiplatform (win32, linux, mac, etc) opensource opengl Battlezone capture the flag game of the same name that you must try at http://BZFlag.org/ or a continual development project with periodic gaming interuptions. See also TimRiker
21:12.28troubledthus min/maxVolunteer arent even active
21:12.34TimRikerinfobot: chanset #infobot minVolunteerLength 50
21:12.34infobotSetting minVolunteerLength for #infobot to '50'; was '1'.
21:12.36TimRikerbzflag?
21:12.36FactBotmethinks bzflag is a 3D internet multiplayer multiplatform (win32, linux, mac, etc) opensource opengl Battlezone capture the flag game of the same name that you must try at http://BZFlag.org/ or a continual development project with periodic gaming interuptions. See also TimRiker
21:12.45TimRikersmacks factbot
21:12.52TimRikerinfobot: chanset #infobot -minVolunteerLength
21:12.52infobotUnsetting minVolunteerLength for #infobot; was '50'.
21:12.56TimRikerbzflag?
21:12.56FactBotbzflag is, like, a 3D internet multiplayer multiplatform (win32, linux, mac, etc) opensource opengl Battlezone capture the flag game of the same name that you must try at http://BZFlag.org/ or a continual development project with periodic gaming interuptions. See also TimRiker
21:13.58TimRikertroubled: I have 2 concerns with that. 1 - it will have unintended concequences. 2 - it's not a required change. just leave the 50 there, or make it larger if you like.
21:16.55troubledwell, ill tweak the numbers in a bit, still gotta test and add all the hooks to return if it shouldnt apply
21:17.59TimRikerconcern 3: adding more things that rely on bot-wide settings is the Wrong Direction.
21:19.25troubledperhaps i can tackle moving stuff from the global conf to chan only stuff another day
21:22.03troubledwell that worked for one test :)
21:22.09troubledlooking good, give me a few
21:35.51troubled[  1197] infobot <TheStaff!i=hyperion@irc/TheStaff> identified
21:35.56troubleddamn, possible bug there
21:36.29troubledminor maybe. seems the bot identified me to the account just by a pm for :INFOBOT query
21:57.15troubledlol -> SELECT factoid_key FROM factoids WHERE factoid_key='bob'
21:57.36troubledlong night tonight I can see
22:00.08troubledanyways, ill probably on and off this for a bit. ill save any sort of commit until tested and working and using some agreeable defaults, l8r
23:41.22troubledwell, think its all done and tested
23:41.31troubledtried all the modes and everything seems great
23:41.53troubledTimRiker: ^^^
23:42.55troubledalso, since i have to hook in Statement.pl to catch learning and check limits when not addressed but in hungry mode, it seems that to allow something like country codes to work, a min of 2 is the most you can have. as for max, i left at 1000 in my bot but its going to need to be somewhat high
23:43.21troubledmaybe 50-100 range I figure should be okay for most stuff
23:43.36troubledanything larger and you better off going manual factoids
23:44.00troubledi still need to track down this spare select sql though. not sure where that sucker is hiding out yet

Generated by irclog2html.pl Modified by Tim Riker to work with infobot.