vendredi 9 décembre 2011
Par Carmen Lotek le vendredi 9 décembre 2011, 11:43
element
a github account. in 2006, github didn't exist, SVN kept breaking in fabulous ways, i'd noticed -git3 suffixes on linux and figured git was worth a try. only host around was repo.cz, which is fitting enough as my grandparents are from neighbouring countries like Slovakia and Poland. i put Boston as my location, although ive never actually lived there, only Canton, Brookline and Quincy. im willing to relocate, so long as your city doesn't require a car. i've strong preference for open-source so if you're talking relocate, that's definitely a requirementpower of limitations
the 80s are an interesting era for dance music, despite incredibly primitive tools like drum-machines limited to 16th-note positions, and 128K of sample RAM to work with - which was a second or two - numerous classics were produced like altered states and entire catalogue of Dance Mania and Mr Fingersthe limitations imposed on element are for few dependencies. if external utilities are used they will be via shell as bindings have historically caused a host of FFI-releated leak/crash issues and are not available on non-C rubies like JRuby or Ruboto. this cuts down significantly on exponential/quadratic likely-breakage as depended-upon libs upgrade or version-pinned deps fail to run on a newer Ruby, both of which i experienced quite a bit due to 1.8-1.9 transition and even more basic issues like rubygems breaking. one of my heroes, drobilla touts his projects as dependency-free. im not quite that extreme and depend on rack.
another prohibition was external database daemons. this was borne out of repeated issues with mysql being upgraded by a package manager, somehow puking on old config files, and needing TLC to restart. and the fact that i dont want to go inserting and spidering the whole disk. a request-time database is populated by translating filesystem data to RDF
applications
instead of blindly stabbing-in-dark, i had various applications i wanted to use, such as general-purpose email/webmail, media-player, feed reader, and faceted-visualizer and these apps and above limitations drove the creation of a framework that would be adaptable to these needstriplestore
a store for RDF is achieved in various ways. either creating files of RDF or non-RDF format - in which case you define a triplizer - or direct inserting and indexing on a filesystem-backed store implemented by the trie structure inherent in pathswebserver
the interface to the storage and triplizing / rendering layer is of course HTTP. even when i'm reading mail on my phone, theres a little server running. this frees from having to work with platform-specific GUI APIsnews / lifestreams
i could cut down on time checking into various tools and apps by integrating everything into a single timeline. it turned out some guy wrote his PhD thesis on this almost 2 decades ago and i'd merely reinvented it. amusingly i saw an email to my boss in SF in one of the screenshots, to chat with him on an article he was writing for WIRED...visualization
around 2005 a group at MIT called Simile was coming up with a variety of data-presentation tools with names like Exhibit and Timeline. i'm quite a fan of these but encountered performance issues with my one intended use-case of Exhibit and it seemed like a good challenge to come up with something fast enough. after banging my head on index structures for a bit and considering using a server-component (choice of newly-funded Exhibit3) i realized CSS could be used to filter the result set. thus began examine an minimalist realization of the core-functionality of Exhibit and eventually other ideas such as timegraph and protovis/d3 alternatives for basics like histograms and finally output modes to serve as input to Protovis and the aforementioned Simile tools in the first placePar Carmen Lotek le mardi 6 décembre 2011, 06:56
Perhaps this is another solution for the Fermi paradox. Societies tolerate innovation until they develop powers frightening enough to make them stop. This point is the opposite of the singularity. Call it 'the exit'. If the exit happens before the singularity, societies may never develop technology sufficient for space exploration on any scale.
this is a good point although i want no part of it, analyzing behaviors and discussions by marketers, security agencies,or hedge funds trying to predict activity is the most likely path to AI, via the indelible need to progress towards reading minds, so hack away..
« billets précédents - page 4 de 73 - billets suivants »

