Par for the course, Also refactoring showcase #1

 Of course, as per usual i neglect to update this even slightly...

i have a handful of drafts, but they never get published.


At any rate, new goal for this blog:
Showcasing neat ways in which refactoring has altered sections of code i'm actually working on.

Too often, people will show "demo code" to highlight how certain techniques are useful, i intend to use this blog to show how i'm refactoring actual production code

First off, two small examples:

Example #1:

This code comes from an algorithm to translate an artificial "genome" into a neural network layer, specifically the part where i decide which "flavor" of network layer to use

Before refactoring:

After:

 

GetActivation checks the available data, as well as validating that an activation function was correctly generated, throwing an exception upon failure instead of returning null.

Example #2:  

Same program, now coming from some genetic mutation-related code, specifically the "mutation" that will randomly select one of the two chromosomes to use within a genome:

Before refactoring:

 

After refactoring(stage 1):

After refactoring (stage 2):


 

A big chunk of duplicated code that got separated out, functionality for adding chromosomes got simplified a lot and extension methods are quite nice

Comments

Post a Comment

Popular posts from this blog

Building SIC-1 in VCB Part 1: RAM

Image
 So i've started a (hopefully) short, and fun side-project again. I'm going to try to emulate the SIC-1 CPU within Virtual circuit board. The SIC-1 CPU is an 8 bit CPU with 256 bytes of addressable RAM i just realized i'm currently making it 256 bits but i'll fix that later (incorrect, turns out i was working on 8*8 bytes) And a single instruction: SUBLEQ A, B, [C] it's a S ingle I nstruction C omputer that should be relatively simple to emulate. As an added challenge, however, i'm also limited to using a Universal gate  to do so,  A universal gate, then, is able to "emulate" other gates, these are the NAND and NOR gates. I've chosen to use the NAND gate, Here's a comparison with their "standard" gates I then added some more designs that might come in handy, a pulse-former seemed useful, and i needed some way to delay signals, an R/S latch also seemed like it'd be needed for RAM, and i also added in a D-flip flop and a manual cloc
Read more

On the intricacies of affordance-based interfaces in C++

Image
Well that's a rather obscure title, isn't it, and it'll require some explanation about some non-standard C++ structures i've been building at work. First, i've done a little bit of work in the GO programming language, and i really enjoyed the mechanics of it's non-inheriting interface polymorphism (i've called them affordance-based interfaces) Next, i found this talk explaining how to do type erasure in C++ in order to essentially emulate these affordance-based interfaces: I wrote a prototype of this system such that i could make a set of affordance classes that describe a specific member function, and might serve as a "delegate" for that function, and an interface class that could combine several of these affordance classes into a single type-erasing interface. This part actually uses a bit of "regular" polymorphism to attain it's goal, but since i've written macros that generate all the required code for you, you can effec
Read more

Building SIC-1 in VCB Part 3: Fadder and auto-RAM

Image
 Last time i finished the RAM, an early design for registers and did some initial trial and error with reading data from RAM via registers Next i began some early ALU work, starting from the "traditional" full-adder i made a while back I translated it into a NAND-only version,  which, after some compression, ended up being the exact same size as the original so i could easily hook it into the same test rig Then i started working on automating the RAM read/write cycle, so i could write with registers G/H and read with F/H (H will become the program counter register) This process was somewhat painful since the RAM read signal is just barely long enough to fully store the data, so the write line has to be active before the data arrives, but stop on the same tick that the data line turns off so there's still some leftover wiring from when i tried to have it re-write its own data (which doesn't work since hysteresis can't easily set its own data when the circuit turns
Read more