Sunday, December 2, 2012

Comparing computer benchmarks

[this post was previously mangled into markup garbage somehow.]

Get the geekbench program--
this will give you some kind of performance number for your laptop/desktop/server:
http://www.primatelabs.com/geekbench/

Also you can view results others have posted here:  http://browser.primatelabs.com/
(search by model, processor type, etc. and view the range of results expected.

This type of metric is much more meaningful for comparisons than 'GigaHertz' -- processor speed.
Going simply by CPU speed is like comparing a sports car to a semi-trailer truck.
Your specific needs will depend on whether you just want to get somewhere fast, or have a lot of stuff to haul.

Sunday, January 22, 2012

tab sweep, 2012.01.22 part 1

I had the idea of doing a tab sweep before I ever heard it called that, but having found this name, I might as well use it.  Sharing is better than not (unless it's crap(), links are the spider silk that holds the web together, and writing a one-line review will help me remember it better than simply adding it to another bookmark folder etc.  And it may inspire some discussion and knowledge sharing, or at least serendipitous clicking...

  • http://www.codemaps.org/ -- apparently has been around for a couple years; there are some diagrams of open source packages like Lucene... okay, on closer look it appears this is mostly(?) advertisement for an eclipse plugin from Architexa that does UML diagramming for mostly java.  Which is fine.  But like other academic spin-offs, it is unclear how much backing and (business) drive behind, balanced against how much more useful it might be if it were given away, and how much more it might be adopted.
    And the related business issues:  "How do we make money if we give it away?"  ".. but then N-times more people will use it, a small portion will result in sales/licensing/training/customization" etc.
    There is some paper on something called 'codemaps' that might be a visual studio plugin..
    As with mindmaps, UML diagrams seem to have a useful life only during part of the learning curve; after you are familiar enough with the terrain, the map is no longer useful.  Or you need a different, showing different information, at different scales.
    I am also interest in 'active' maps -- I'm not sure I've seen a UML diagram displaying runtime activity or stats, except maybe special purpose in some academic papers.
    For UML diagramming while learning code / class relationships and similarities,
    ArgoUML is pretty nice.  Read in your source files and then you can drag-n-drop classes of interest and it automatically adds the relationships.  Hmm..  how to handle updates?
    UMLGraph is free and can be automated; there is an ant task that can add diagrams to your javadoc too, but since it's really a diagram specification, and GraphViz handles the drawing, it is not tied to any language or tools or formats..

    I like the concept of being able to 'visualize software' but haven't seen the magic bullet yet..
  • Tracking the Trackers: Where Everybody Knows Your Username from Stanford Law School
    not particularly new for me, because I have been at one of the top online ad companies, and took part in design of a project that would receive user info from the ISP (income, demographic profiles) and join it with other information providers than had credit scores, household buying info, etc.
    There are a number of links and tools there to follow, and also a well-described methodology.  I am somewhat inspired against to work on a rule-based rewriting proxy that can persist my browsing history or be use for spidering / heuristic search, and also answer / collect info on topics like
     - how many sites use X (where X is some package, framework, software version, hosting provider, technology feature, etc.)
     - what ad networks does this site use, and the reverse query
     - what is this cookie format?  how is it decoded?  and, (if multiple users shared this) how does it vary per user?  can it be hacked?
     - how often is this (html) tag actually used?  who uses it (a package, a tool generator, a design firm)? 
     - trend analysis (when did this term first appear / reach some threshold), latent semantic tools
     - general IR capabilities, because information is easier to retrieve when you have information.

    Something like muffin (a customizable proxy with plugin architecture), db storage, a simple rule base handles a lot of this..
.. that was only like 2.5 tabs I had; to be continued.