2020-12-16
1346
#vanilla javascript
Victor Jonah
30678
Dec 16, 2020 ⋅ 4 min read

What is railway oriented programming?

Victor Jonah I am a Software Developer with over three years of experience working with JavaScript and its frameworks. I currently work as a remote software developer for a tech agency.

Recent posts:

leveraging Lighthouse audits to optimize web performance

Leveraging Lighthouse audits to optimize web performance

Slow-loading pages can stem from multiple causes, which makes them one of the most challenging issues to fix in web development. Lighthouse can help you detect and solve your web performance issues.

Anna Monus
May 14, 2025 ⋅ 6 min read

Building multi-region infrastructure with AWS

This isn’t theory. It’s the exact setup you need to deliver fast, resilient apps across AWS regions with zero fluff.

Marie Starck
May 13, 2025 ⋅ 5 min read
the nine best FaunaDB alternatives for 2025

The 9 best FaunaDB alternatives for 2025

Looking for a FaunaDB alternative to migrate to? Examine nine other platforms you can use and factors to consider when choosing an alternative.

Nefe Emadamerho-Atori
May 13, 2025 ⋅ 7 min read
Techniques To Circulate And Record Knowledge In Engineering Teams

Techniques to circulate and record knowledge in engineering teams

From onboarding to bug tracking, these knowledge-sharing techniques keep your team aligned, reduce overhead, and build long-term technical clarity.

Marie Starck
May 12, 2025 ⋅ 4 min read
View all posts

5 Replies to "What is railway oriented programming?"

  1. In the last example, wouldn’t you still need to check isValidated before calling sendMail(), or maybe passing it in would be more in line with this design pattern? Or maybe I’m missing something, its late. 🙂

  2. Its is really unclear how the last example allows for decomposition of a single failure/sucess value to multiple firstname, lastname values

  3. Am I missing something? This seems like a basic overview of the Maybe Monad with respect to exceptions, but without any of the work shown. You haven’t given any mechanisms for ‘self-healing’, or dealing with the unhappy path at all, just that you CAN switch to one and that it’s better to write small functions than to throw all error handling into the main function.

  4. The principles of this “railroad” concept is nothing new. I was taught error handling like this over 30 years ago.

    One minor suggestion – rather than “success” or “failure”, the reality is that there is success, recoverable failure, non-recoverable unit failure, and non-recoverable app failure. The developer needs to determine if a failure is recoverable (e.g. login failure, connection timeout, etc.) or non-recoverable. If the latter, the developer needs to decide if the app is now unusable, or just one portion of it.

    Another suggestion – don’t just return success or failure. Include the appropriate enum flag, but also include in the return object details about what went wrong, such as all the error messages in the exception stack, snapshot values of runtime variables, what module, method, and line number the error occurred, etc., not only for logging, but for useful information for the user in a recoverable failure.

Leave a Reply