IRC log of fenfire on 2005-03-02

Timestamps are in UTC.

16:35:13 [fflogbot]
fflogbot has joined #fenfire
16:35:13 []
topic is: - the channel may or may not be logged
16:35:13 []
Users on #fenfire: fflogbot ffdarcsbot CaptSolo antont humppake tuukkah ibid mudyc majukati benja_
16:35:19 [tuukkah]
if the bot writes to a file in realtime, the file can't end in the xml close tag so it can't be an xml document
16:35:40 [benja_]
tuukkah: they're re-created or something
16:35:52 [benja_]
they're different per day, so the load isn't so big
16:36:10 [benja_]
16:36:11 [tuukkah]
hmm :-(
16:36:19 [benja_]
"" -- argh
16:36:39 [benja_]
need to change this
16:36:48 [tuukkah]
16:36:58 [antont]
16:38:34 [tuukkah]
and I thought the html file was a lot more verbose than plain text... :-)
16:49:46 [fflogbot]
fflogbot has joined #fenfire
16:49:46 []
topic is: - the channel may or may not be logged
16:49:46 []
Users on #fenfire: fflogbot ffdarcsbot CaptSolo antont humppake tuukkah ibid mudyc majukati benja_
16:51:27 [benja_]
16:51:32 [benja_]
now it's barely working
16:51:57 [benja_]
no index files yet
16:52:45 [tuukkah]
so, I could load the nodes but how to see them?
16:53:22 [benja_]
16:53:23 [tuukkah]
oh, right...
16:53:31 [tuukkah]
let me try
16:53:56 [benja_]
they only use dc:description which is not a text property, so it's not very rewarding yet ;-)
16:55:03 [antont]
what's rdf:li ?
16:55:15 [antont]
not 'list item' like in html i suppose..
16:55:31 [antont]
oh it is in <rdf:Seq> so perhaps yes
16:55:45 [tuukkah]
after I went to irc:// I could see something but ran instantly out of memory
16:56:07 [benja_]
tuukkah: huh, very weird
16:56:10 [benja_]
there isn't a lot of data
16:56:21 [benja_]
well, MX=256M =-)
16:56:50 [benja_]
ah, I know
16:56:58 [benja_]
there's some loop so the DFS bites =)
16:57:18 [benja_]
you can set maxdepth to a lower number in StructureView
16:57:30 [tuukkah]
or increase heap
16:57:45 [benja_]
tuukkah: that was the MX comment
16:57:57 [tuukkah]
16:58:36 [tuukkah]
oh, that was a makefile variable
17:01:25 [tuukkah]
argh, all those rdf:_38 on the property list =)
17:02:29 [benja_]
17:03:10 [tuukkah]
trying to got to eventlist node caused recursion limit to be exceeded
17:03:38 [tuukkah]
at org.nongnu.libvob.impl.awt.DepthSorter.quicksort(
17:04:26 [benja_]
17:07:06 [tuukkah]
with smaller limits it works
17:07:21 [tuukkah]
maxdepth 4, maxring 3
17:08:18 [tuukkah]
it doesn't really matter so much if dc:description isn't showed as node content
17:08:41 [tuukkah]
worse is that rdf_n aren't sorted
17:09:23 [tuukkah]
there's a lot to tweak...
17:09:52 [tuukkah]
the different events from same nick aren't known to be from the same person
17:11:55 [benja_]
yeah, I don't understand why they aren't sorted, will look into it sometime
17:12:16 [benja_]
it does matter to me, the text should be in the node
17:12:47 [tuukkah]
how would you sort them anyway? alphabetical isn't enough with rdf:_3 and rdf:_21
17:12:50 [benja_]
tuukkah: true ... hard to see whether they should be
17:13:10 [tuukkah]
matters, but not so much :-)
17:13:13 [benja_]
since nicks aren't permanently associated with people. hmm
17:13:34 [tuukkah]
first of all, freenode has registered nicks
17:13:38 [benja_]
tuukkah: you would sort them by the number after the _ I suppose ;-)
17:13:56 [benja_]
tuukkah: true...
17:14:12 [tuukkah]
second, as long as you don't quit and join, the person can't change
17:15:21 [antont]
except when you change the nicks, but it's quite uncommon, and also communicated to the channel
17:15:48 [antont]
i agree that the speaker should be 'known', whatever that means
17:15:52 [benja_]
antont: but not in a way that would help RDF =)
17:16:28 [tuukkah]
so instead of "some benja_", you can have "benja_ who has registered the nick" or "the user who joined the channel on 2005-02-17T18:54:09, and is currently knows as benja"
17:16:40 [benja_]
the bot could record the mbox_sha1sum if mbox is available, but not everybody has it set...
17:16:59 [tuukkah]
why does it have to be mbox_sha1sum?
17:17:06 [benja_]
tuukkah: ?
17:17:30 [tuukkah]
why can't it be anything that identifies the person?
17:17:54 [benja_]
tuukkah: well, because the bot can discover it
17:18:04 [benja_]
homepage is the one other thing it could discover
17:18:14 [tuukkah]
how could it discover those?
17:18:15 [benja_]
I think
17:18:21 [benja_]
tuukkah: you ask NickServ
17:18:31 [tuukkah]
and how can't it discover if the nick is registered?
17:18:43 [benja_]
tuukkah: ?
17:18:58 [tuukkah]
<benja_> antont: but not in a way that would help RDF =)
17:18:58 [benja_]
it can discover if the nick is registered and has these set
17:19:27 [benja_]
tuukkah: 19:15 < antont> except when you change the nicks, but it's quite uncommon, and also communicated to the channel
17:19:28 [tuukkah]
if it's registered then that is enough to connect the different log entries
17:19:37 [benja_]
saying "I changed my nick" isn't helpful for RDF
17:20:10 [tuukkah]
but the bot can understand the message and mark it in rdf in some more direct way
17:20:17 [benja_]
tuukkah: so you want to make a new vocabulary with registeredFreenodeNick?
17:20:27 [benja_]
of course these can still change owners
17:20:48 [tuukkah]
why not?
17:21:08 [tuukkah]
I can enter your email address in my nickserv records?
17:21:22 [benja_]
tuukkah: what are you replying to?
17:21:27 [antont]
if you want more spam, that is?
17:22:04 [tuukkah]
or homepage url, if we're more against spam
17:22:23 [antont]
17:22:26 [tuukkah]
benja_, why not have attribute registeredFreenodeNick?
17:23:07 [benja_]
19:20 < benja_> tuukkah: so you want to make a new vocabulary with
17:23:07 [benja_]
17:23:08 [tuukkah]
benja_, and if I have your contact info in my nickserv records, it doesn't make me you
17:23:15 [benja_]
I wasn't saying you couldn't make one?
17:23:37 [benja_]
tuukkah: it doesn't, but what's your point?
17:23:50 [tuukkah]
isn't "why not" an appropriate way to say "yes" in this case?
17:24:39 [tuukkah]
benja_, using email address from nickserv records has some problems too
17:24:44 [tuukkah]
nothing else
17:24:52 [antont]
but i'm off - cu. it was pretty interesting to see structured irc data. and i didn't even manage see that rdf-xml as very horrible. and with the improvements discussed seems nice. although i've no idea about the applications ppl have in mind .. but even just being able to link to arguments in discussions is useful i guess
17:24:55 [benja_]
tuukkah: way to say yes: no, IMHO ;-)
17:25:15 [benja_]
I didn't understand it as a "yes, I want to make a new vocab"
17:25:43 [benja_]
tuukkah: ok, it does, but it's a general trust problem...
17:25:47 [benja_]
17:26:12 [tuukkah]
perhaps I don't want to. but why couldn't I... :-) I thought perhaps you had some convincing arguments, as you probably do :-)
17:26:33 [benja_]
registeredFreenodeNick is a possibility, but of course they *can* change owners, we'd just ignore that problem
17:26:42 [benja_]
tuukkah: =)
17:27:28 [tuukkah]
is there anything that can't change owners ?-)
17:28:34 [benja_]
social security numbers? =)
17:29:01 [benja_]
well, it's true that foaf:mbox ignores the problem too
17:30:26 [tuukkah]
could you do the old trick and somehow add a timestamp to the id, be it mbox or nick?
17:31:16 [benja_]
there's foaf's holdsAccount --> OnlineChatAccount
17:31:32 [benja_]
tuukkah: no, that would require picking one
17:31:44 [benja_]
which has the same problem as picking a URI
17:31:50 [benja_]
how do you get people to agree on one?
17:32:15 [tuukkah]
I don't see the problem
17:32:18 [benja_]
because if you don't, you lose what you want to get: being able to know that different nodes identify the same resource
17:33:29 [benja_]
if I use and you use, the computer can't infer that the two are the same
17:33:37 [benja_]
i.e., that we're talking about the same person
17:34:03 [tuukkah]
true, but if I used 2004-01 as you've told me
17:34:42 [tuukkah]
basicly, you could tell that your mbox is
17:35:21 [benja_]
tuukkah: that's not a realistic proposal for use today, because most people have not picked such a date
17:35:36 [benja_]
if they did, they could just as well pick a URI to represent themselves
17:35:58 [benja_]
and we didn't need to identify them by mbox
17:36:01 [tuukkah]
now someone else starts to use the box, they would stick with
17:36:14 [tuukkah]
17:37:56 [tuukkah]
you can have several mboxes, right, so the dated one would be a more strict reference to you and the dateless one a less strict
17:38:51 [tuukkah]
well, the trust just isn't there yet in the systems
17:39:36 [benja_]
tuukkah: a reasonable workaround at the moment is to trust everything unless there's manual intervention =-)
17:39:56 [benja_]
that's what all the semweb stuff does today
17:40:55 [tuukkah]
at the moment I can publish FOAF that says my mbox_sha1sum is same as yours :-/ would any software be able to let the user say anything agains this?
17:41:30 [benja_]
tuukkah: depends on who you consider the user =)
17:41:33 [tuukkah]
that is, is manual intervention implemented in any software
17:41:42 [benja_]
fenfire would allow the user not to load your foaf, for example ;-)
17:42:01 [benja_]
foaf aggregators apparently usually allow the admins to exclude files
17:42:14 [benja_]
I've asked about this and have been told it has happened, but rarely
17:42:20 [benja_]
(and they fixed it manually)
17:43:02 [tuukkah]
17:43:14 [benja_]
it's an active research topic of course
17:43:32 [tuukkah]
as long as it's a quad graph, you can filter out the offensive source
17:43:54 [tuukkah]
but I'm afraid you couldn't undo smushing, for example
17:45:20 [benja_]
tuukkah: in Fenfire, you could
17:45:29 [benja_]
in most smushing impls, you couldn't
17:45:32 [benja_]
17:45:49 [benja_]
I decided this was important to ff because the user could accidentally smush
17:46:10 [tuukkah]
ok, this is very nice to hear
17:46:40 [tuukkah]
(I did another ctrl-g on the irclog, and the update seems to have worked)
17:46:48 [benja_]
i.e., the user mistypes mbox='', deletes 'fi' and types 'com'
17:47:13 [benja_]
if '' also exists in the system, the two nodes and all info about them would've become smushed
17:47:29 [benja_]
but SmushedQuadsGraph will unsmush when the reason for smushing goes away
17:48:31 [tuukkah]
so the smushing isn't ever permanent?
17:48:41 [benja_]
hm, what do you think about just making it foo:freenodeNick for now -- i.e. not change registration? (much easier to do I believe)
17:49:17 [tuukkah]
what do you mean "not change registration"?
17:49:21 [benja_]
tuukkah: the unsmushed quads graph is always maintained and the relationship to the smushed graph is functional if there are no bugs
17:49:42 [benja_]
s/not change registration/not check registration/, sorry
17:50:32 [tuukkah]
it's not difficult to check registration, is it? at least on protocol level, each PRIVMSG is marked whether it's by a registered nick or not
17:50:55 [benja_]
tuukkah: but adding freenodeNick likely involves no coding at all
17:51:16 [benja_]
checking registration likely involves some perl and Net::IRC coding =-)
17:51:24 [benja_]
(you're welcome to do it of course ;-))
17:51:56 [tuukkah]
ok, it's not like this is security sensitive
17:52:16 [tuukkah]
I know a perl hacker who is going to attend the course :-)
17:53:52 [benja_]
the only way it would affect anything outside the logs is if you put the freenodeNick in your foaf
17:53:58 [benja_]
or some other public location
17:54:48 [tuukkah]
and if you do that, I think everybody understands that irc nicks can be stolen
17:55:29 [benja_]
tuukkah: the only way it would cause a real problem is if you dropped your nick, somebody else picked it up, and put it in *their* foaf
17:55:37 [benja_]
and that's extremely unlikely
17:55:46 [benja_]
plus you'd remove it from your foaf most likely
18:03:43 [tuukkah]
so, would you add dc:description to Fenfire.TEXT_PROPS?
18:03:52 [benja_]
18:04:04 [benja_]
I believe it's not really appropriate here
18:04:34 [benja_]
rdfs:label would be more appropriate I think
18:04:43 [benja_]
18:05:03 [tuukkah]
shouldn't there be two things anyway: a text that someone says and type of message like privmsg, notice, join, part
18:06:05 [tuukkah]
or would that be rdf:type
18:06:13 [benja_]
well, it's just the text that someone said
18:06:16 [benja_]
at the moment
18:06:20 [tuukkah]
which is foaf:chatEvent now
18:07:01 [tuukkah]
it's not strictly saying as stuff like title is a chatEvent as well
18:07:24 [benja_]
well, ok
18:07:52 [benja_]
I think making the model there more complex is unnecessary work at the moment
18:08:05 [tuukkah]
18:08:19 [benja_]
I'm not opposed to it, just have no interest in doing it
18:08:28 [tuukkah]
18:08:52 [tuukkah]
would all dc:dates be shown in the calendar view?
18:09:03 [benja_]
ah, right, crschmidt mentioned
18:09:10 [benja_]
tuukkah: y
18:09:15 [benja_]
well, for now
18:09:21 [benja_]
in the future, we'd have selection by type
18:10:10 [tuukkah]
how do i see the calendar view?
18:10:21 [benja_]
when the user can create a new vocabulary with dates inside Fenfire, and make it an applitude with a specific calendar view in an integrated process, then we'll really have "fenfire"
18:10:34 [benja_]
tuukkah: go on something with a date, hit Ctrl-V
18:10:50 [benja_]
it may not be performant
18:10:57 [benja_]
you have been warned
18:10:59 [benja_]
18:11:00 [tuukkah]
18:11:15 [benja_]
I need to go shopping, promised
18:11:18 [benja_]
(shops close at eight)
18:11:19 [benja_]
18:11:21 [tuukkah]
but this would be more useful than sorting the stupid property names
18:11:30 [tuukkah]
18:12:07 [benja_]
tuukkah: a) sorting the stupid property names is important because many stupid rdf vocabs use containers ;) ;)
18:12:48 [benja_]
b) I'd prefer to sort by date; dajobe said that different items can have the same date and order can be important (multiline paste)
18:13:09 [benja_]
I'm thinking about having date + number-of-item-with-same-date
18:13:18 [benja_]
I really dislike the use of a container
18:13:43 [benja_]
it's ugly that there's a different container for every day and stuff...
18:13:57 [benja_]
but, cu
19:04:47 [benja_]
shops lo
19:04:52 [benja_]
shops closed, benja_ back
19:07:42 [tuukkah]
that was quick :-)
19:08:25 [benja_]
btw, if you don't know, have a look
19:08:49 [benja_]
(a community making video series with 5min per episode and voting on them)
19:10:33 [benja_]
(try, e.g., the first episode of The 'Bu @ -- I use noatun to watch them)
19:16:15 [tuukkah]
umm, a truly funny squirrel
19:16:42 [tuukkah]
unfortunately I don't think I have red-blue glasses nearby ;-)
19:17:05 [benja_]
I did try it with glasses the second time, I think the images come from a single cam actually =)
19:18:44 [tuukkah]
yes, I was wondering if they'd really had a stereo camera
19:19:07 [benja_]
I thought they might've used two, don't know if that's really possible
19:20:50 [benja_]
so now we have the magic potion, Lob/Model templates, and Tuomas' Functional system
19:21:20 [tuukkah]
the languages?
19:21:26 [benja_]
19:21:48 [tuukkah]
what's up next =)
19:21:58 [tuukkah]
turn it all into common lisp %-)
19:22:03 [benja_]
19:22:33 [benja_]
19:22:44 [tuukkah]
or haskell that can output either Java and use AWT or C++ and use OpenGL
19:23:04 [benja_]
lobs are an object system, really, but the template system is used as a kind of programming language
19:23:43 [benja_]
models are something of a cross between an object and a function, or set of functions
19:23:47 [benja_]
21:23 < benja_> lobs are an object system, really, but the template system is
19:23:48 [benja_]
19:23:51 [benja_]
used as a kind of programming language
19:24:51 [benja_]
and then, lobs and models and graphs are observable
19:25:18 [benja_]
there must be a way to simplify all this
19:25:25 [benja_]
separate it into orthogonal parts
19:25:41 [benja_]
in a way, a template is a function returning an object
19:26:37 [benja_]
templates use parameter lists with named parameters, potion uses positional parameter lists, the functional stuff uses one parameter per function
19:28:00 [tuukkah]
those all can be represented as one parameter that can be a dictionary with indexes as keys :-)
19:29:01 [benja_]
which reminds me: functions and lobs have some constraints w.r.t. speed
19:29:16 [benja_]
perhaps potions does too, when mixed with the two above
19:29:30 [benja_]
i.e., when we use potions to specify views (view parameters)
19:30:41 [benja_]
I failed an exam in "fundamentals of theoretical computer science"
19:31:38 [tuukkah]
:-( graphs, logic, type theory?
19:31:58 [benja_]
no. computability, turing machines, P/NP, languages
19:32:12 [tuukkah]
oh, that stuff :-)
19:32:20 [benja_]
I don't think I would have failed graphs, logic, type theory ;-)
19:32:54 [tuukkah]
where these two classes track A and track B?
19:33:18 [tuukkah]
majukati has documented some really funny procmail hackery
19:36:14 [benja_]
tuukkah: they have nothing to do with this course
19:36:37 [tuukkah]
neither a nor b?
19:36:39 [majukati]
tuukkah: yep, that was awesome.
19:37:15 [tuukkah]
don't the message headers break procmail reading the message as .procmailrc?
19:38:14 [majukati]
fflogbot you are illegal to be run from eppi. we had only right to run archive changes commit bot
19:41:35 [majukati]
benja_: you need to use another host
19:42:20 [benja_]
majukati: I recall the conversation differently, I believe we explicitly talked about logging
19:42:46 [majukati]
we explicitly talked about logging and the answer was no
19:42:55 [benja_]
so what was the plan?
19:43:10 [benja_]
since they don't want us to run a web server on siksak