Bayram Eker
Menu
EN
CV Start a conversation

Home / Writing / Practice

What I choose to share as open source

Sharing a repository starts a relationship with its readers. Here is how I choose between releasing code, explaining research and publishing an architecture.

TaskNebula coordination platform
TaskNebula is one of the projects whose source can be inspected and used. Product interface.

I think about the first hour someone spends with a repository: reading the instructions, installing the dependencies and trying to understand what is actually there. Publishing the code is an invitation to spend that hour. I want the invitation to be honest about what the person will find and how much help I can offer afterwards.

TaskNebula and QANTIS have MIT-licensed editions. NowFlow Community and qmesh use Apache-2.0. NeuraOS shares architecture documentation, without distributing its source or binaries. These are different ways of making work available, and I want a visitor to be able to tell which one they are looking at.

Start with what a reader can actually use

For QANTIS, the public repository is an explanatory research edition. It lets someone examine the approach and follow its structure. It does not contain the full experimental tooling. Saying that clearly is more useful than leaving a researcher to discover the missing pieces after an afternoon of setup.

Maintenance is part of that choice. A small edition with clear instructions can be useful for years. A larger release that I cannot support may leave someone dependent on a tool they cannot repair. I would rather make the scope understandable at the start.

Different products have different boundaries

NowFlow Community shares the workflow engine and agent patterns. Its commercial edition adds the integration surface, with 188 blocks and over 300 connections, alongside operational support. NeuraOS takes another approach: public runtime and governance architecture, with a proprietary delivery path from signed model to device.

NODERIQ Defense has no open edition. Its public architectural explanation and a distributable implementation are different things. Across these projects, I try to make that distinction easy to see before anyone invests time in using the material.

Why sharing is still worth the effort

I want someone reading a paper to be able to inspect how its approach is expressed. I also want other developers to find mistakes or uses I have missed. TaskNebula adds a more direct reason: a team can use something helpful without first becoming a customer.

Research software may only reach a small audience. A repository star cannot tell me whether someone understood a paper because of the code, or used one part to get their own project moving. I keep public figures dated, but those smaller possibilities are why sharing remains worth the effort to me. They begin with a person opening the repository and finding what the page promised.

Three ways to make work available
  • Code to run and change
  • Research to follow and question
  • Architecture to understand the design

Written by Bayram Yüksel Eker. Corrections and disagreement are welcome at bayram@neuraparse.com.