IRC log of fenfire on 2005-03-26

Timestamps are in UTC.

09:20:50 [tuukkah]
based on the recent slashdot article, I wonder how google does version control and releases on their unified codebase
10:22:06 [ibid]
11:30:44 [tuukkah]
it sounded a lot like fenfire: they have a lot of code that is divided into subprojects and anyone can commit to anything
11:31:03 [ibid]
11:31:13 [ibid]
* ibid does not read /. frequently:)
11:31:41 [tuukkah]
11:49:16 [benja_]
11:52:43 [ibid]
11:55:00 [tuukkah]
huomenta benja_
11:57:07 [ibid]
benja_: i don't know if you noticed, but i finally got fenfeed running for the first time :)
11:57:17 [ibid]
benja_: it helped when i realized i don't have to package it first... :)
11:57:54 [benja_]
11:58:02 [benja_]
I hadn't yet noticed
11:59:06 [ibid]
do you know what causes the extreme slowness between the creation of the window and the showing of the ui components?
11:59:23 [ibid]
doesn't sound like anything i'd guess
11:59:47 [benja_]
no, but I expect it to either have to do with threading problems or with smushing
11:59:57 [ibid]
11:59:59 [benja_]
I haven't investigated
12:00:24 [ibid]
any guess where in the code, roughly?
12:00:28 [benja_]
threading problems: the feeds are loaded on a different thread, then inserted into the main graph in a bg task on the main thread
12:00:43 [benja_]
this would be in org.fenfire.fenfeed :-)
12:01:28 [benja_]
smushing: the current smushing impl is very non-optimized, but I don't have evidence it's a problem, yet; the code would be in org.fenfire.swamp.smush.SmushedQuadsGraph
12:01:46 [ibid]
i think i've been explained before what smushing is but can't remember
12:01:47 [benja_]
it would probably pay to profile
12:02:03 [ibid]
* ibid might try to review the code later today
12:02:08 [ibid]
no promises :)
12:02:16 [benja_]
ibid: making two bnodes (in our case URIs too) that represent the same resource into one
12:02:48 [ibid]
what's a bnode?
12:03:34 [benja_]
a node without a uri. is graph-local. also known as anonymous node
12:03:40 [benja_]
bnode = blank node
12:03:42 [benja_]
(no label)
12:03:45 [ibid]
12:03:48 [ibid]
thanks :)
12:24:28 [rubberpaw]
rubberpaw has joined #fenfire
12:25:07 [rubberpaw]
* rubberpaw checks to see if anyone is awake
12:26:21 [tuukkah]
hi rubberpaw. at least benja_, ibid and me :-)
12:26:26 [rubberpaw]
Hi tuukkah
12:26:36 [rubberpaw]
I have a bit of a favor to ask, if it's possible
12:26:49 [rubberpaw]
I'm just finishing up a paper for the ACM hypertext 2005
12:26:58 [tuukkah]
at least you can ask
12:27:00 [rubberpaw]
and I want to see what happens if I put a particular dataset into Fenfire
12:27:09 [rubberpaw]
(I was thinking about the screenshow I saw with the gnowsis stuff)
12:27:16 [rubberpaw]
12:27:21 [rubberpaw]
screenshot, that is
12:27:36 [rubberpaw]
(kb issues through this ssh tty, sorry)
12:27:45 [tuukkah]
do you have the rdf file?
12:28:02 [rubberpaw]
and I didn't know if anyone had a build I might be able to run on OS/X
12:28:47 [tuukkah]
I think benja_ has OS/X
12:28:52 [rubberpaw]
no RDF at this point, though I could probably whip it up. It might take a bit. I have a 60 node, geodesically-structured hypertext
12:29:50 [rubberpaw]
I think Fenfire might be my best opportunity to find software that properly visualizes it
12:30:27 [rubberpaw]
gZigZag does fairly well in MindSundew view, but the issue of directional links causes problems.
12:31:19 [tuukkah]
hmm, what kind of problems? because rdf triples are directional too
12:32:58 [rubberpaw]
well, gzz links are posward or negward
12:34:10 [ibid]
actually, gzz links are both posward and negward :)
12:34:16 [rubberpaw]
when setting up the links in gzz, I used one dimension's posward and negward links to show links around pentagonal centers
12:34:29 [rubberpaw]
ibid: yeah
12:34:54 [rubberpaw]
which worked great
12:34:58 [ibid]
* ibid is writing a biting mail on identity to the course list :)
12:35:55 [rubberpaw]
but then I created another dimension to store the third link for a triangle.
12:36:41 [rubberpaw]
This worked well, but when I linked pentagons to each other, moving outside of the pentagon would be either left or right -- I couldn't make it uniform
12:36:52 [rubberpaw]
argh -- it's difficult to describe.
12:37:07 [rubberpaw]
let me try again.
12:37:16 [tuukkah]
I think I get the picture
12:37:39 [rubberpaw]
my geodesic structure contains 12 groups of 5 triangles
12:37:48 [rubberpaw]
I made each of these a loop of nodes
12:37:54 [rubberpaw]
in one dimension.
12:37:57 [tuukkah]
12:38:18 [rubberpaw]
now, each of these triangles also has one outward link to another pentagonal loop
12:38:38 [rubberpaw]
I wanted each of these links to be in the same direction.
12:39:03 [rubberpaw]
ie, it would create something like a tree that would branch downward forever.
12:39:58 [tuukkah]
do you know how you would represent this in rdf?
12:40:19 [rubberpaw]
but the posward/negward-ness of links meant that it was impossible to tell which direction one would have to move.
12:40:34 [rubberpaw]
Actually -- not at this point. I haven't put much thought into it.
12:41:08 [tuukkah]
or as a general graph?
12:41:23 [rubberpaw]
12:42:07 [rubberpaw]
no, I have not. I am, unfortunately deficient in that part of mathematics.
12:42:19 [tuukkah]
basicly, you have 60 nodes, and some lines between them
12:42:46 [rubberpaw]
right. I have already mapped it out in hypertext software.
12:43:24 [tuukkah]
it's a general graph, when the lines don't have direction or colour or anything like that
12:43:25 [rubberpaw]
2d spatial hypertext
12:43:31 [rubberpaw]
ahh, ok.
12:44:19 [tuukkah]
with zzstructure, you have to choose colours and arrowheads, and you can't have to arrows of same color be of same direction to one node
12:44:51 [rubberpaw]
12:45:21 [tuukkah]
with rdf, you can choose colours, you have to choose arrowheads but you can always put arrow in both directions, as you don't have limit on the number of arrows in a node
12:45:57 [rubberpaw]
right -- in zigzag you can have unlimited arrows, but only two per dimension.
12:47:21 [tuukkah]
on some level of rdf, you can describe that some relation is of a type where the direction doesn't matter, but fenfire doesn't make use of such description at the moment
12:48:08 [tuukkah]
at least if we're discussing the fentwine structure view
12:49:23 [rubberpaw]
am I correct in thinking that if something links to or is linked from a the currently-focused node, Fenfire will show it in the view?
12:49:34 [tuukkah]
my point is, I think in practise you'd want to choose a direction instead of making two-way connections
12:50:07 [rubberpaw]
perhaps for the visualization, yes
12:50:23 [rubberpaw]
in the actual hypertext (, direction is meaningless.
12:50:33 [tuukkah]
rubberpaw, yes. links to the node are on the left half of the wheel, and links from the node are on the right half of the wheel
12:51:05 [rubberpaw]
that's what I need. Since the hypertext is spherical, this will look natural.
12:56:41 [rubberpaw]
a couple of screenshots can be seen at:
12:59:21 [rubberpaw]
for interface purposes, I suppose I could modify my gzz file to have two links between the pentagon edges
12:59:31 [tuukkah]
oh, you've managed to put it quite nicely into MindSundew! I wonder if it's even possible to get to look anything like that in current fetwine
13:00:06 [rubberpaw]
but it only looks good up to a certain point. Because if I go down to any of the other pentagons
13:00:22 [rubberpaw]
it messes up, since the link back to the pentagon currently viewed is negwawwrd
13:00:36 [tuukkah]
oh, right
13:00:46 [rubberpaw]
so, I would be happy to settle for only seeing one node
13:00:49 [rubberpaw]
and three surrounding nodes
13:01:08 [tuukkah]
we hope that it would at some point be easy to make your own views
13:01:16 [rubberpaw]
which is what I imagine Fenfire would be able to do.
13:04:23 [tuukkah]
ok, now I have a loop of five linked nodes in fenfire :-)
13:05:24 [tuukkah]
I suppose I should write a script that creates the rdf data so that it could be faster to experiment with alternatives
13:07:11 [rubberpaw]
wow -- if you really feel like it :-).
13:08:11 [tuukkah]
I feel like I haven't learned the rdf structure and the wheel view in the same way as I did gzigzag 0.6 at some point
13:08:20 [tuukkah]
so this could help
13:08:46 [rubberpaw]
13:09:13 [rubberpaw]
*works to procure a screenshot of gzz mindsundew when it doesn't look so neat.
13:10:45 [rubberpaw]
* rubberpaw hompes he still has the dataset -- it was four months ago...
13:10:54 [rubberpaw]
(argh. kb -- hopes)
13:18:36 [rubberpaw]
whew. Good.
13:44:56 [ibid]
tuukkah, benja_: see my mail?:)
13:47:32 [tuukkah]
sorry, too long, I'm hacking ;-)
13:47:56 [ibid]
the answer is yes, then :)
13:48:33 [ibid]
i think i'll reprint that as a blog posting; any objections? :)
13:53:29 [rubberpaw]
rubberpaw has quit
13:55:07 [rubberpaw]
rubberpaw has joined #fenfire
13:55:17 [rubberpaw]
* rubberpaw is rebuilding the data so each trans-pentagon link consists of two links in gzz -- just to see if that fixes the visualization issues
13:55:25 [rubberpaw]
whoops -- disconnected
13:56:27 [tuukkah]
I have the pentagons now, and I'm thinking how to connected them to each other in the right way
13:58:36 [rubberpaw]
oh cool.
14:09:16 [tuukkah]
this is what I have at the moment:
14:10:48 [ibid]
14:14:43 [rubberpaw]
hmm. So it's showing more than just the immediately linked items
14:15:10 [tuukkah]
there are two parameters for controlling that in the view
14:15:33 [rubberpaw]
that's the idea though. The reality is that the pentagons on the sculpture
14:15:54 [rubberpaw]
are not supposed to be identified as groupings. Each triangle is the center of it's own three-linked universe
14:16:02 [rubberpaw]
or at least, that's the goal.
14:17:12 [tuukkah]
one is how many links to take when starting from the focused node. the other is how foggy nodes are still drawn
14:17:52 [rubberpaw]
ahh, ok. So if I got a copy of fenfire and the dataset, I could tweak things?
14:19:22 [tuukkah]
actually, the only way to change the parameters at the moment is in the source code
14:19:36 [rubberpaw]
14:20:09 [rubberpaw]
* rubberpaw grumbles and considers whether he has time to install apple Dev tools
14:20:18 [rubberpaw]
hmm, I'll finish the recoding in
14:20:38 [rubberpaw]
gzigzag and then see where I stand. Thank you very much for your efforts.
14:21:00 [tuukkah]
it was nothing
14:21:15 [rubberpaw]
do you mind sending me the dataset?
14:21:39 [tuukkah]
the rdf file?
14:21:39 [rubberpaw]
* rubberpaw has a copy of the source, but just had a crash and is still restoring software packages.
14:21:42 [rubberpaw]
14:21:53 [tuukkah]
14:22:58 [rubberpaw]
I'll send you my gzigzag folder when I finish tweaking it.
14:26:29 [tuukkah]
hmh, the parameters really don't work as I'd expect. I can get results such as no links showing at all, though
14:29:12 [rubberpaw]
14:30:22 [rubberpaw]
btw, when I make two links between trans-pentagon nodes, spreadsheet-type views look nice.
14:33:01 [tuukkah]
ok, I thought this totally wrong =) it looks right when I set maxring to 0 and maxdepth to a big number like 10
14:34:28 [rubberpaw]
14:37:24 [tuukkah]
btw, if you have the code from benja_, you could ask him to give you an http account so you can pull updates made to the unpublished code
14:37:48 [rubberpaw]
ahh, ok. will do.
14:38:21 [rubberpaw]
I was using the source from the site.
14:38:48 [rubberpaw]
interestingly, after I reached a certain point, entering data became very easy (now that I have the double link)
14:39:04 [rubberpaw]
since I have a spatial model of the sphere in front of me, finding a particular node with the kb is very easy.
14:39:15 [rubberpaw]
14:39:25 [tuukkah]
yes :-)
14:39:57 [rubberpaw]
I don't have to use the screen even, since I know how the structure works.
14:43:13 [rubberpaw]
hmm, it's behaving much more nicely in the mindsundew now
14:43:46 [rubberpaw]
I think I have it all linked properly.
14:46:32 [rubberpaw]
yeah. .zips the files...
14:47:14 [rubberpaw]
13k -- should I upload or email?
14:49:07 [tuukkah]
upload would be easier for me
14:49:59 [rubberpaw]
ok. I'll put it on my server in a few minutes...
14:51:58 [rubberpaw]
whoops -- I misnamed my dimensions. I called trans-pentagon dim "clockwise" and the pentagon links "transagon"
14:52:24 [rubberpaw]
just keep that in mind when you view.
14:59:39 [rubberpaw]
15:02:17 [tuukkah]
403 Forbidden
15:03:29 [rubberpaw]
oh ak. let me change the perms...
15:03:45 [rubberpaw]
try now
15:03:55 [tuukkah]
15:04:09 [rubberpaw]
15:07:54 [tuukkah]
the animation became a bit funny =)
15:08:08 [rubberpaw]
15:08:10 [rubberpaw]
but it works?
15:08:24 [tuukkah]
15:08:43 [rubberpaw]
15:08:57 [rubberpaw]
that is because the node immediately to the right
15:09:01 [rubberpaw]
also appears below
15:09:31 [rubberpaw]
and below that, the top pentagon is reproduced at the very bottom
15:09:44 [rubberpaw]
this is a result of the double-linking along the "clockwise" dimension
15:11:16 [rubberpaw]
argh. I would have fixed the naming error, but for some reason, renaming the dimension caused links to break.
15:15:03 [rubberpaw]
Well, I'm including you in the acknowledgements, if you don't mind.
15:15:36 [tuukkah]
it's ok if yoy want to
15:15:41 [tuukkah]
15:20:59 [tuukkah]
rubberpaw, I think I was able to change the dimensions. from the dataset directory, I renamed the file "transagon" as "clockwise" and vice versa
15:27:33 [rubberpaw]
ahh, ok.
15:27:46 [rubberpaw]
it still works? (I wasn't aware of how things worked on the file end)
15:27:53 [tuukkah]
15:28:12 [tuukkah]
not very user-friendly way to achieve renaming, thougn ;-)
15:28:59 [rubberpaw]
hehee. Can you email it to me?
15:29:39 [tuukkah]
I'm thinking wouldn't it be easier if you rename the files yourself?
15:30:44 [rubberpaw]
lol. This is what I get for having three monitors, and three desks of papers strewn about. A lack of focus.
15:32:10 [tuukkah]
I have to go to the store before they close - I'll be back later
15:34:23 [rubberpaw]
ok. ttyl
16:07:01 [ibid]
16:15:33 [rubberpaw]
rubberpaw has quit
17:03:38 [rubberpaw]
rubberpaw has joined #fenfire
17:03:52 [rubberpaw]
*sigh* keeps disconnecting
17:05:29 [ibid]
reason: Read error: 60 (Operation timed out)
17:05:33 [ibid]
on both occasions
17:06:00 [rubberpaw]
weird. I have a good connection. It must be something about my client
17:06:10 [rubberpaw]
thanks ibid
17:06:54 [rubberpaw]
BitchX, but you already know that.
17:09:57 [ibid]
google gives me nothing
17:12:24 [rubberpaw]
oh well. Thanks
17:12:39 [tuukkah]
17:12:53 [tuukkah]
I looked at another channel and same error message there
17:14:07 [tuukkah]
but he has a bad connection
17:14:24 [rubberpaw]
oh well.
17:16:17 [tuukkah]
internet is so complex
17:16:30 [rubberpaw]
17:17:48 [tuukkah]
there must be a lot of attention already paid on simplicity, but I feel a lot more could be done
17:18:30 [rubberpaw]
to make the 'net better?
17:19:06 [tuukkah]
would robust be the word?
17:19:23 [rubberpaw]
ahh. Well, it's pretty tough to do it.
17:19:34 [rubberpaw]
I mean, it's supposed to in theory be very robust with packet switching
17:19:47 [tuukkah]
on that level, yes
17:19:49 [rubberpaw]
but there are so many single points of failure
17:20:02 [rubberpaw]
because it's wired by economic forces
17:20:30 [tuukkah]
they always tell how internet is meant to be the ultimate P2P without points of failure :-)
17:20:30 [rubberpaw]
I mean, in our area, Comcast had everything going through one piece of hardware
17:20:50 [rubberpaw]
and when that piece of hw went down
17:20:57 [rubberpaw]
half of the nodes on the net on the US east coast went down.
17:21:05 [rubberpaw]
this was just last month
17:21:32 [rubberpaw]
what we need is wireless with GNURadio
17:21:34 [tuukkah]
probably because one piece of hardware was cheaper to buy and manage than several
17:21:46 [rubberpaw]
or they were lazy
17:21:51 [rubberpaw]
17:21:57 [rubberpaw]
they needed to surveil the connection
17:22:08 [tuukkah]
right :-/
17:22:40 [rubberpaw]
but if we all used auto-negotiating wireless devices
17:22:46 [rubberpaw]
in an ad-hoc network
17:22:55 [rubberpaw]
then we could leverage the robustness of the protocols
17:23:18 [rubberpaw]
but the gov't and big biz wouldn't allow it here
17:23:27 [tuukkah]
but what about this irc server giving error messages that in addition to "read error", 60 and "operation timed out" would tell who and why does it matter :-)
17:23:42 [rubberpaw]
one of my profs built a supercomputer in his basement.
17:24:03 [rubberpaw]
and a week later, the FCC came to his door, since it was interfering with television signals
17:24:08 [rubberpaw]
this was decades ago
17:24:08 [tuukkah]
rubberpaw, yeah, I've heard of the issue of radio spectrum monopoly
17:24:42 [rubberpaw]
(well, it's sticky. I would have to maybe give the IRC my email, since the timeout could mean I lost connection)
17:25:08 [rubberpaw]
(and maybe I want to be anonymous)
17:25:23 [rubberpaw]
(but yeah -- the error message could be clearer)
17:25:48 [tuukkah]
but we'd want to know on which layer the trouble might lay
17:26:08 [rubberpaw]
(right. Well. The server is OSS. Get the source and patch ;-) )
17:26:20 [tuukkah]
17:26:24 [rubberpaw]
if the connection goes down
17:26:34 [rubberpaw]
perhaps you can get the irc server to remember my ip
17:26:40 [rubberpaw]
and try to traceroute me
17:26:48 [rubberpaw]
noting the point at which the connection dies
17:26:53 [rubberpaw]
and reporting it
17:27:08 [tuukkah]
that would be bonus :-)
17:27:45 [rubberpaw]
but you're right. These things should be built into basic internet libs
17:27:59 [rubberpaw]
there should be a high level diagnostic system that will try to identify issues like this
17:28:15 [rubberpaw]
and let the programmer decide who/how to notify
17:28:29 [rubberpaw]
you just basically write some really good error handlers.
17:28:54 [tuukkah]
I know I could start with Fenfire ;-)
17:28:57 [rubberpaw]
that can be overrided for particular actions but get passed a package of error data
17:29:09 [rubberpaw]
there might already be a lib
17:29:17 [rubberpaw]
I bet that it would be really useful for Storm
17:30:49 [tuukkah]
but we have even lots of simple things still like catching exceptions and turning them into something visible instead of proceeding quietly
17:31:34 [rubberpaw]
in Storm?
17:32:18 [tuukkah]
probably there as well, but I'm thinking of the apps, FenFeed and Fentwine
17:32:30 [rubberpaw]
ahh, ok.
17:33:02 [rubberpaw]
17:33:43 [tuukkah]
in Storm, I guess we should start that kind of checks with getting through NATs and firewalls
17:34:00 [tuukkah]
as they are used everywhere these days
17:34:18 [rubberpaw]
17:46:26 [tuukkah]
ibid, I wouldn't call urn-5 "the brainchild of the Fenfire project" but "a brainchild ..." instead ;-)
17:46:39 [ibid]
17:46:59 [ibid]
fixed in blog
17:49:50 [ibid]
sort-dctrl works!
17:49:56 [tuukkah]
I've been reading a lot of grammar lately, although not English. 2 points for each such failure, and 2 in every sentence would fail you
17:50:20 [ibid]
in what kind of a course?
17:50:38 [ibid]
of course that would not be a submission in any course where that matters that much :)
17:51:00 [tuukkah]
grammar in swedish basic studies
17:51:12 [ibid]
in general terms, most of my emails and virtually all of my blog postings are "first draft" material, with little checks
17:51:40 [ibid]
if grammar was an issue, i'd certainly have read it a couple more times over
17:51:54 [tuukkah]
I didn't want to say you should check more
17:52:56 [ibid]
i didn't think you did :)
19:02:25 [lilo]
[Global Notice] Hi all. Sorry to disturb. Freenode is setting up an interim incident reporting page, and we need your help. If you have ideas for *specific* categories of problems to be reported or support requests to be made, please stop by #freenode-tracking and, erm, blurt them out on the channel? Remember, please show good netizenship and don't join just to be joining. Thanks! :)
19:19:04 [rubberpaw]
rubberpaw has quit
20:39:57 [rubberpaw]
rubberpaw has joined #fenfire