I’ve been using the i3 Window Manager for a little while now on my laptop. For those of you who don’t know, i3 is a tiling window manager somewhere between wmii and awesome.

I personally find it’s minimal enough to be quite nice on my 15" laptop, but still implements useful things like notifications, systray, and a normal way to log out.

My customizations are available on my github with config file an support scripts. The big things it does is set up i3lock (with a fuzzed background, thanks to Image Magick), starts a bunch of baseline stuff, and fixes the Chrome App List.

So keyboard-based window management is here to stay for me. Good luck getting me on vim or emacs, though.

I’m working on a new project which I thought I’d turn into a series. It’s nothing fancy or sophisticated. The code name is Leatherbound, and it’s a diary/journal web app. This first post I just want to lay out the premise and some of the requirements and guidelines.

• Easy to use
• Responsive design

Some of the decisions I’ve made:

• Using Material design: Because why not? At least it’s not more bootstrap)
• Google AppEngine: scalable and free limits
• Google Cloud Storage: I think NoSQL will work better than SQL
• Python: Obviously
• Google Login
• Using Polymer (although that might not be mobile friendly? May ditch it)

Some features:

• Optionally include events that have occurred since the last post (from Google Calendar)
• Full WYSIWYG (I’m thinking inspired by hallo.js, but probably re-implemented for material)
• Embedding media, including photos

Some other random details:

• First screen after login is the new entry screen, and it should load fast. I figure its the #1 thing people do when they login.
• May look into Ember? The initial load time is a concern against the previous requirement and Polymer’s already slow loading.

Next time, I’ll be showing the prototype interface I’ve got going so far.

(See all the posts with the #leatherbound tag.)

Seeing as I’m a maker, I’d have two ideas how to do this:

1. Tear apart an off-the-shelf voice changer and exchange the speaker or other components for your own to get the behavior you need.
2. Build your own voice changer

Either way, check what others have done (Hack A Day).

Keep in mind good speakers just take space. You can’t avoid this.

I thought I would formally announce that I have changed companies. I’m now in a position of management in a software as a service company. If you want specifics, I’m sure you can find out.

I know I haven’t been posting much, but this is Important. Gnome needs help.

Even if you don’t personally use the Gnome desktop environment, you have certainly benefited from their work: GTK (used by Gimp, Pidgin, and others), practically every Linux distro has something they developed or promoted (gconf->dconf, D-Bus, etc).

Please donate!

Allow me to summarize my experience with patent searches:

1. Try to come up with as many ways to describe various aspects of your product
2. Search Google for these descriptions
3. Freak out when none of them resemble what you’re doing.

To future employers: If you like me and want to keep me, do this.

I don’t have an experience with this, but it resonates true with me. Not a great metric, I know, but it’s often the best I got.

If you know better, leave a comment below!

About five years ago, I bought a cushy couch for my office. (Okay, yes, I did get the model that could flatten into an emergency nap station, but let’s just say that I plan for contingencies—it sounds more professional that way.) Our projects required a lot of office-to-office visiting to discuss situations in person, and eventually, said couch (and therefore, my office) became a veritable beacon, attracting anyone looking for an excuse to decompress. Such is the life of a one-couch, 50-chair business.

This has 0 technical content. But it’s an important aspect of people. I guess it’s just a reminder that everyone has inner demons and struggles with something. Nobody is all strong all the time.

I have an idea for how to process the raw image:

1. Use the calibration information in each photo to make a model of the lenses
2. Create a normal map of the raw image show where each ray came from
3. Create a color map of the raw image for future demosaicing
4. Project rays from the modeled CCD and ray trace it to your selected focal plane
5. Apply some filter to demosaic and rasterize

I still need to process for a depth map. Maybe look for where rays converge?

Tonight, I shall be giving a talk at the local Python Users Group on metaclasses and related. If there are any questions, feel free to ask them here!