Towards a unified digital aproach to knitting

Our idea is to create a knitting library for a format that allows conversion of knitting projects, patterns and tutorials. Usually, communities will only focus on the knitting format for their machines. Our approach should be different and be able to support any knitting communities efforts.

Here is our strategy to achieve this:

  • We connect to different communities to get a broader view on what their needs are.
  • Our knitting format is based on knitting instructions like knit, purl, yarn over, skp. We found a comprehensive list on Wikipedia.

Other Communities

From time to time we meet with other people who also knit and could use our software.

First, we met with Viktoria from ETIB Berlin. She taught us a lot about knitting, how she does it, that almost everything could be created from one peace with the machine. Also, that AYAB is used for lace patterns. We saw examples where she let meshes fall so that larger holes were created. Our goal is to support laces in the file format.  Color patterns should be possible across sewing edges.

We are also in touch with Valentina Project. With their software we would be able to connect to yet another community and use their sewing patterns for custom-fit clothes.

We got in touch with Kniterate. They and we share a lot of goals. Because they create a startup, they are very cautious what they release. They focus on their open-source knitting machine first and later on the software. They already created an editor much like we imagined ours to be, but as a web application. A way of collaboration could be that we understand their file format and see how we can support it.

Only talking about our GSoC project is worth it as other people may have seen alike at Maker Faires and other hacky places. We have the chance to bring communities and efforts together.

Knitting Format

A universal knitting format has many concerns:

  • Languages of users differ
  • It should be possible to knit by hand
  • Mesh sizes and wool differ
  • Different knitting machines with different abilities
  • A knitting format for exchange is never complete. A knitting format for machines must be complete.

In contrast to a knitting format for a automatic machine, it is possible, to have machines operate in semi-automatic modes or just to knit by hand. In both cases, meshes could be changed in a way that was never foreseen. This is why we did not base it on meshes and mesh types but rather on instructions – closer to the mental model of the knitters who perform instructions with their hand.

Some of the instructions are understood by the machines, some could be adapted a bit so the machine can do it automatically or faster and some are still necessary to be done by hand. We created a Python module for that, “knittingpattern“. We work on it in a test-driven way.


Exception handler in KnitLib

An exception is an error that happens during the execution of a program. It is important to have exception handler in KnitLib to handle and deal with errors automatically. Many standard libraries define their own exceptions to report errors that may occur in functions they define.  Depending on the kind of error in knitlib (“communication error”, “pattern not found error”, etc.. ). The exception handler module in the knitlib can handle the exception and the knitlib can be continued afterwards with the previously saved data.

Define exceptions as classes

Exception can be defined as classes which do any other class can do, but are usually kept simple, often only offering a number of attributes that allow information about the error to be extracted by handlers for the exception. When creating a library like KnitLib which can raise several distinct errors, a common practice is to create a base class for exceptions and subclass that to create specific exception classes for different error conditions.

Part of the Exception Handler Module, catch exceptions raise from the library

Importance of the test cases for the KnitLib

Having test cases is very important especially for a library like KnitLib because using test cases; we can clearly test particular fields.  In KnitLib, test cases show the information of how the KnitLib should be checked. Also test cases help for new contributors to understand about the KnitLib.

There are several test cases for the current KnitLib implementation such as tests on ayab communication, tests on ayab image, tests on command line interface, tests on KnitPat module and tests on knitting plugin.

For an example in ayab communication there are several important functions have been tested. Test on closing serial port communication, test on opening serial port with a baud rate of 115200 which ayab fits, tests on sending start message to the controller, tests on sending line of data via serial port and tests on reading line from serial communication.  Most of these tests have been done using mock tests. Mock is a python library to test in python.  Using mocks we can replace parts of our system with mock objects and have assertions about how they have been used. We can easily represent some complex objects without having to manually set up stubs as mock objects during a test.

It is very important to improve further test cases on the KnitLib because with the help of good test cases we can guarantee that the KnitLib’s features and functionalities should be working great.



Knitting machine abstractions for Knitlib

Hello, during the last weeks we have been working on Knitlib and Knitpat, a knitting machine control library and a standardized format that allows for exchange and storage of patterns.

In order to achieve a common platform for knitting machine development we have the need to abstract away implementation details that can difficult the generic usage of the lib, while keeping extensible and powerful control features. Among the most important abstractions developed for Knitlib is the Knitting Machine Finite State Machine, an abstract representation of the procedures needed to operate a knitting machine.

BaseKnittingPlugin, the basis of knitlib’s machine knitting controller plugins.

The architecture of Knitlib allows for easy integration of different knitting machine plugins for varied use cases, hardware, and software protocols. All functions of the plugin are non blocking except for .knit(), which is blocking due to the physical interaction needed in order to execute this command. To ease usage and to enable more versatile behaviour from the .knit() function, without limiting the interaction facilities needed for operation, the callback infrastructure allows for blocking and non-blocking callbacks from the Plugin to the machine operator (the Knitlib client), such as Information, Warnings, Error Notifications or Mechanical Required Actions (moving spools, switches, needles, etc). Callbacks abstract away the notification and interaction paradigms from the plugin, allowing for simpler behaviour, a more elegant design and ease of testing. Callbacks also allow for future plugins to not take care into implementing user interfaces, but to focus on functionality.

The pending remaining challenge is to standardize configuration options, flags and settings in order to allow for UI that respond to each plugin requirements and options, and to specify which features are supported on each machine plugin. Insofar, most of the standardization has been done on Knitpat, but some specifications such as physical resource assignation (Serial Ports, input streams, etc) are still to be implemented soon.

Thank you, and I hope this article helps you to understand the software architecture and design patterns of the implementation of Knitlib.



Knitting Library to support numerous Knitting Machines

Hi everyone, I’m Shiluka Dharmasena, Computer Science and Engineering undergraduate from University of Moratuwa, Sri Lanka. Thanks very much for this great opportunity to get involves in FOSSASIA open source project and I’m keen to give my fullest contribution to the FOSSASIA.  For GSoC 2015, I am developing a library to support knitting machines. It’s great to work with FOSSASIA team under awesome mentors Mario Behling and Christian Obersteiner. This is the Initial suggestions for the project.

Library to support knitting machines

Basic abstraction of the library to create library as separate layers

  • Serial communication : open, configure, read and write to serial port
  • Machine definitions : machine initialization, load to machine, save from machine
  • Image handling functions : memory allocation, read image file, setter and getter for image pixels
  • File handling functions : read / write access
  • Common utility functions

Add dependencies for the library

There are several dependency libraries which ayab has such as pillow, pyserial, wsgiref, fysom and yapsy with relevant versions. These dependencies should be added to the library.

Functions for the library

For access hardware directly
* Identify existing libraries to interface with hardware
* Use libraries to deal with serial ports
* Ability to get all the functions via library (for an example functions in and add more functions to support in more machines)
* Ability to send QT signals

Emulate file formats

* Create functions to emulate file format

Add Arduino support


Installation should be well documented in step by step. Additionally I suggest to add GUI based installation with usability improvements with standard installation procedures.

* Installation on Linux (32bit, 64bit)
* Prerequisites
* Setup
* Installation on Windows (32bit, 64bit)
* Prerequisites
* Setup

Tests on the library

Use a python test framework like unittest to test the python library. It supports test automation and aggregation of the tests into collection.

Documentation of the library

Documenting Class definitions, methods definitions and links to the given source codes.