The week I spent out of the fray weighs heavily, and it’ll take a while for the backlog to unjam.

Slogging through entirely too much e-mail and having a few somewhat critical things to unknot that spilled over into my personal time turned the week into a blur of hopping around and cradling my laptop until late in the evening.

A few decisions were made, most news was barely glossed over, a bunch of unfinished notes and to-dos were added to the pile.

First off, I decided to quit learning R, probably for good. It’s trendy and suitably hyped and Big Data-ish, but the unfamiliar syntax and libraries were hampering me, so I installed IPython and got to grips with matplotlib and pandas instead.

And believe me, it made a world of difference. Being able to invoke web services and have the results rendered straight into an IPython notebook is stupefyingly amazing, all the more so considering the notebook interface works very nicely indeed (even if it’s browser-based) and plays very nicely with Dropbox-like storage1.

Plus the learning curve was pretty much nil, except for the finer points of pandas DataFrames. And I haven’t even started on the natural language stuff I’ve been pondering, or playing with the d3.js plugins.

It feels oddly like going back to Mathematica, really, although I’m not using it for any real math. But I’m tempted to play around with its built-in distributed processing and eventually bolt it onto Hadoop somehow – although that too has had a far steeper learning curve than I expected, and to this day I wish there was a better way to work on it (cleaner abstractions, better language support, etc.)

On a somewhat unrelated twist, I’ve been reading up on Solr for a project we’re kicking off (wanna work at SAPO? See here, and it’s nice to see how far we’ve come since Lucene (and derivatives like Whoosh, which I use in Yaki). The thing’s amazingly fast, very well thought out, and a dream to work with2 for what I have in mind.

On the tech news front, well… The things that stuck were the PS4 vaporware announcements (going x86 and supporting game streaming when there’s an opposing trend of going ARM and downloadable strikes me as anachronistic), the Ubuntu “we’re not Android but we’re building off CyanogenMod” anomaly and, of course, glass as a fashion statement – which will likely never work properly with my prescription.

Can’t say I’ve missed much.

  1. The notebooks are stored as JSON files with embedded data (source code and graphics), and thus sync speedily to all my machines. ↩︎

  2. Once you’ve got batch posting figured out, since its JSON API subscribes to the notion that you can have multiple items with the same key – which is anathema to me. ↩︎

See Also: