Ask HN: What steps to take to rebuild an existing application? | Hacker News Reader

Ask HN: What steps to take to rebuild an existing application?

8 pointsladybro6 comments

I've been tasked with rebuilding a successful desktop application into a web application. A few relevant points:

- The app is in the scheduling realm and involves some semi-complicated algorithms

- I did not contribute in any way to the desktop app and it's written in an old language that I don't know at all

- I plan to build the new app in Ruby on Rails, which I am experienced with

- We've identified ~75% of the current functionality that needs to be included in the first version of the web app

- The desktop app has had strong acquisition interest in the past and the best case scenario for all parties would to be acquired in the next 2 years

What steps would you suggest I take to make the process as successful and pain-free as possible?

View on HN

(1) If you can recycle the old "semi-complicated algorithms" in some way you should. For instance if these are written in FORTRAN or C it is not that hard to write wrapper code to access them in a more complex language on the back end, it also may be able to compile them to webassembly.

(2) If you can't do that, or even if you can, you should build a test suite that works on both the old and the new code so your reimplementation is correct. This will save you so much suffering.

(3) If you want to make an app that has the same sensibility as a desktop app in 2020 I would look to React, Vue, Svelte or something like that.

If you want it as pain-free as possible, optimize for the transition once it's ready, and so avoid making improvements over the UI. Leave UI improvements until after people are using it.

I've been working on rebuilding a legacy system for the last "few" years, and now that it's mostly ready, we have a nightmare in figuring out how to transition the users as seamlessly as possible. They can do everything, simpler, but it's different, and will require training. Improvements would have been a lot better to do piecemeal after they've started using the new codebase. After discussing it a few months ago, we came to the agreement that we really should have done a port that resembled the original as closely as possible to ease the transition.

Too late to back out of it now, though.

As others already mentioned, the best thing would be if it is possible to reuse the non-UI code (so logic & algorithms) and put it on a server behind an API.

On the other hand, if the code is not written in the modular fashion and things are tightly coupled, it might be harder than writing everything from scratch (of course, that depends a lot on the volume of the code).

As PaulHole suggested, writing a test suite would also be very important to make sure your code works as expected - that will save you a lot of trouble in the long run.

RoR seems like a good and stable choice for the backend, do you plan to use something like React/Vue/Angular on the frontend?

On the side note, I am working on a project that would hopefully make app dev easier and more sustainable in the long run, but unfortunately it is still in the very early stage of development. Would appreciate your feedback though! https://wasp-lang.dev/

I think systems like WASP will proliferate. React, Vue, Svelte and similar frameworks have gotten people used to code generation, transpiling, etc. Technically, people are ready for something that blurs the front-end/back-end distinction.

Thank you very much for your feedback PaulHoule! Feedback like this motivates us to keep going :)

Dump the desktop applications logic layer into a Web Api and build the Web Application off of that, basically just re-implement the UI

Reply on news.ycombinator.com