Threads for doriantaylor

  1. 2

    I tried to get the IBIS webapp mentioned in the first video running. I apparently succeeded, despite really not knowing anything about perl. This is more or less what I was able to muddle out (I may have forgotten a step):

    git clone
    cd p5-app-ibis
    # This adds 'cpanm' which something said to use.
    cpan App::cpanminus
    cpanm inc::Module::Install
    cpanm Module::Install::Catalyst
    perl ./Makefile.PL
    cpanm Convert::Color
    cpanm RDF::Trine
    # At this point I noticed that ./Makefile.PL made a Makefile, but it sure didn't help with the dependency hell.
    cpanm Data::GUID::Any
    cpanm Encode::Base58::BigInt Data::UUID::NCName
    # This sure didn't do it.
    cpanm --installdeps .
    # This package has a test failure.
    cpanm --force Data::UUID::NCName
    cpanm MooseX::Types::Moose
    vi app_ibis.conf
        change line 66
            dsn       dbi:Pg:dbname=trine
            dsn       dbi:SQLite:dbname=trine
    1. 2

      Brave of you to try to fire it up. I probably haven’t updated the Makefile.PL in a while, thus the missing dependencies.

      I made that thing in 2013 for the express purpose of testing RDF::KV. It turned out to be marginally useful so I kept poking at it for year or two afterward, but consider it to be way too sclerotic for what I actually need out of a tool like that. I’m in the process of rewriting it (including a completely different visualization).

      1. 1

        Replying real quick before (the time switch of) my modem turns itself off for the night (v. personal form of regulating an ADHD-related thing):

        Extremely cool that you’re actually trying it out, you are a webizen after my own heart. I’m following your instructions right now, see if I can get Ibis running, too.

        While make runs, some initial thoughts on playing around with the prototype at :

        • the non-labeled, non-spatiality of the circle graph gives me no sense of place/structure/what connects to what. really, the only thing it tells me how many/few relations (incoming/outgoing/looping) a node (Issue/Position/Argument) has.
        • the compact list on the front page of every Issue, Position, and Argument in the reasoning, even without information? That is awesome to get a feel of ‘what sort of stuff is in here’. It’s like a word cloud but good.
        • okay I have more to type but only minutes left, sending now. Busy few days ahead, might get back to you in a few days? [turned out I was just too slow, briefly turning on mobile Internet to send this]
        1. 2

          Yeah the visualization is trash; I just implemented it because it was easy. I have a new one in mind for the rewrite.

          1. 1

            Eh, it may be trash for transferring the graph structure, but it was pretty good at transferring a ‘vibe’. It made a very striking image, made me want to understand it, made me want to know more about this IBIS thing. And I enjoyed trying to puzzle it simultaneously with reading through the issues/positions/etc.

            Also, I am amused to note that after a day or so of not thinking about the image, when I recalled it just now it my brain went “Ah, the one that was like a Star Trek UI, but in a circle!” Make of that what you will. (Maybe because of the round(ed) bits + the colour scheme?)

            1. 1

              Ha, thanks for being so charitable!

              I just cribbed the design (though I wrote my own implementation) from Krzywinksi’s Circos plot because my original design tried to use his hive plot, but what I found was the aspect ratio of the hive plot was unpredictable so I swapped it out for the Circos, which stays put in that regard.

              I’d say the drawback to either is that they aren’t the best for representing the IBIS structure which ends up being pretty (but rarely strictly) hierarchical, so a Sugiyama (GraphViz) style treatment would ultimately be better, which is the general direction I’m headed (with some pretty hefty customizations).

              Another experiment I had going on with this tool was to use the embedded RDFa to marshal the CSS: I use Sass (I can’t remember if it’s computed on the fly or not) to match the palette to RDF classes and properties, which, being embedded in the markup hierarchy, turns out to be a pretty powerful way to style web pages. (Not to mention SVG; I use this technique pretty much everywhere now.)

              I did a more expansive writeup which talks about the history of IBIS and the direction I plan to (eventually) take the tool.