Hi Al,

Thanks for creating pyautogui and for writing the Automate the Boring stuff book.. I saw your youtube video at Pycon 2016/2017, which introduced me to pyautogui and have been using it a lot to automate some stuff.

I have been coding for a few months, working on a side project trying to automate a lot of things that are pretty mundane but need to be done manually. It has gradually got pretty complex, and now I have intermittent bugs, which I can’t necessarily track down. I have been reading Code Complete and gradually implementing more Object oriented programming structure, writing better specifications, and deciding the architecture before I start coding, thinking of writing unit-tests. Because of the intermittent bugs, which I can’t track down, somebody suggested that I put my windows application in a Docker container.

Questions

It seems to me that all of the above strategiest: OO, writing unit-tests, specification, requirements, source control, Docker/pyenve/virtualenv are designed as a defense mechanism against complexity of a program. So we can work with the smallest chunk.

I know that my code base can be improved a lot, but now that I know there are a lot of strategies, that can be use to improved the code base, but I know necessarily know how to prioritize my time so that I can improve my codebase most efficiently. If you were given an unknown codebase, which was partially OO, didn’t have unit-tests, didn’t have requirements/specifications, didn’t use Docker/pyenv, how would you go about prioritizing which one of these strategies to use to improve the code base?

I also have some plans to extend the code base urgently to some other activities, but because of the intermittent bugs, I am wondering if its better instead to stop and make the current code base more OO, have unit-tests, write out specifications, use some kind of environmental variable control/Docker, before I write new code?

What is the best practice in such scenarios?

Are you reffering to this post? https://blog.codinghorror.com/we-are-typists-first-programmers-second/

Hi Al,

Thanks for creating pyautogui and for writing the Automate the Boring stuff book.. I saw your youtube video at Pycon 2016/2017, which introduced me to pyautogui and have been using it a lot to automate some stuff.

I have been coding for a few months, working on a side project trying to automate a lot of things that are pretty mundane but need to be done manually. It has gradually got pretty complex, and now I have intermittent bugs, which I can’t necessarily track down. I have been reading Code Complete and gradually implementing more Object oriented programming structure, writing better specifications, and deciding the architecture before I start coding, thinking of writing unit-tests. Because of the intermittent bugs, which I can’t track down, somebody suggested that I put my windows application in a Docker container.

Questions

It seems to me that all of the above strategies: OO, writing unit-tests, specification, requirements, source control, Docker/pyenv/virtualenv are designed as a defense mechanism against complexity of a program. So we can work with the smallest chunk.

I know that my code base can be improved a lot using the above strategies. But I dont necessarily know which one of these I should implement first.

1. Say you were given an unknown codebase, which was partially OO, didn’t have unit-tests, didn’t have requirements/specifications, didn’t use Docker/pyenv. How would you think about prioritizing which one of these strategies to use to improve the code base?

2. I also have some plans to extend the codebase urgently to some other activities, but because of the intermittent bugs, I am wondering if its better instead to stop and make the current code base more OO, have unit-tests, write out specifications, use some kind of environmental variable control/Docker, before I write new code. In your opinion, is it better to make the current code base more robust using the above strategies, before writing new code?

Hi Alpha..

Thanks for sharing your advice.

Yeah.. my development environment doesnt change too much, but I do my development on my own computer, which turns out to be Windows.

I will work on adding unit-tests and making the new code modular, add unit tests.

Hi Al,

Appreciate your answer and for sharing your experience. I will work on incorporating unit-tests for the existing codebase and new codebase. And work on refactoring once I get some breathing room.

The Tetris analogy to technical debt was very helpful.