New Series
Starting Today,

I’ll be starting a series explaining my current research project. Working in an embedded systemsparadigm, I’ll be investigating hardware accelerators for use in biomedical, wireless, or real time systems.
Read more
Starting Today,

I’ll be starting a series explaining my current research project. Working in an embedded systemsparadigm, I’ll be investigating hardware accelerators for use in biomedical, wireless, or real time systems.
Read moreTake this

Connect it to this

Adjust it with this
And Package it all up cleanly and without any added bulk whatsoever, into this

I would buy the shit out of this product.
My current project at work has been diverted for a few days to work on a c language parser. No, i’m not rebuilding my own compiler or pre-optimizing code for anything groundbreaking like that, I’m analyzing long form C/C++ code for numbers of complex and floating point arithmetic: multiplies, divides, adds etc. I decided to use python since it has a fairly easy regex library as well as nice string manipulation, but this is still a more challenging project than I’d expected.
Here’s the catch, I have a fairly limited amount of a priori knowledge about lexical parsing as well as variable tokenizing, issues dealing with functional scoping, regex and so on. This sounded like something I could mock up in a few hours , a day tops, and i’m surprised by the amount of time it’s taking just to get variable token tables filled correctly.
My current method is to do a first run through the file creating a dict token table with python where the variable names are keys and the data type is their value, also in this pass I’m looking for where the arithmetic functions are, dumping the unnecessary asterisk’s ( *int foo; for example ).
My “fast and loose” solution was to scan for lines with an * initially, then split each word/item in the line and check if it was type complex<float> else just return false for the line. This fell through almost immediately when I captured 1 of probably 30 or 40 complex multiplies in this file.
Next Step:
I suppose the next step will be filtering the content of lines which have an asterisk, throw away the pointers and properly split these lines to get the variables by themselves, as I may be facing a lookup issue in addition to handling scope and control-flow cases, ( a complex multiply in a loop that runs 30 times should count as 30 multiplies, for obvious reasons )
That’s High Praise..
I’m currently working on a submission for the I2P Idea to Product engineering proposal competition. I’m working with a friend on (hopefully) a natural language processing library using genetic algorithms to concise down the codebase from the standard hidden markov model.
Although it’s mostly theoretical and a pipe dream at this point, it would be a fun project to chase down a little bit.
Tonight I wound up with a bit more free time than usual, and needed something to focus on to get my mind off other things, so I finally got around to writing up the Instagram Challenge I’d seen around for a little while. This is my code for the project as well as the image I chose to unshred, the program works fairly well, I added a bit in there to handle edge detection fairly well given that the image in question isn’t a 360 panorama shot, which gave me a bit of trouble.
I haven’t finished the automated number of strip detection part yet, if there’s time i’ll try and finish that up tomorrow.
These are the before after I used to test the program

You might have heard this song used incredibly well in the final scene of this week’s How To Make It In America. I’ve been playing it on repeat for a while now, don’t you just want it to be the soundtrack of your life?
new m83 sounds good
Hmm, now i want cookies