Pagwin/new_blog
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

finished retrospective

Browse source
This commit is contained in:
Pagwin 2025-07-15 16:28:19 -04:00
parent f9876e3eb9
commit e5d0f4dfc7
1 changed files with 16 additions and 82 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

98
posts/soft-eng-project-retro.md
View file

@ -3,9 +3,9 @@ title: "Retrospective on my software engineering class"
description: "Writing about how I felt my software engineering class went"
date: "2025-07-05"
date: "2025-07-15"
draft: true
draft: false
tags: []
---
@ -16,17 +16,17 @@ In short I got an A in the class (somehow) but I failed at my objective(s).
## Objective(s)
My objective starting off with this project was to remake [BinghamonBetterBus](https://git.pagwin.xyz/Pagwin/BinghamonBetterBus) (yes that's a typo, no I'm not fixing it in the repo name lol) with better routing an UI with the assistance of a competent friend (same friend who I was [arguing with about enums](/posts/rust_enums)).
My objective starting off with this project was to remake [BinghamtonBetterBus](https://git.pagwin.xyz/Pagwin/BinghamtonBetterBus) with better routing and UI with the assistance of a competent friend (same friend who I was [arguing with about enums](/posts/rust_enums)).
## Outcome
The ultimate result was something with slightly better routing and a slightly better UI.
Technically this would meet the aforementioned requirement(s) if not for the fact that the original BinghamonBetterBus was made in a 24 hour hackathon by just me while this was done by me + competent friend over the course of over 2 months.
Even compensating for the school work load, i.e. hackathon took place in February when I was in my third year of college and this was From March to May in my fourth year, this is an abysmal showing.
Even compensating for the school work load, i.e. hackathon took place in February when I was in my third year of college and this was from March to May in my fourth year, this is an abysmal showing.
## What went wrong
## What Went Wrong
Two things
Two things:
1) I suck at motivating myself and this class was too lax to compensate for that
2) I underestimated how much I would need to take charge to get this done to the desired degree
@ -34,90 +34,24 @@ Two things
The second point originated from expecting the competent friend to be an equal in the project but not factoring in how busy they were though I did take more charge of things over time.
To expand on the first point though, this class had us set our own weekly objectives to complete and I consistently set the objectives lower than needed to get things done to a satisfactory degree due to a combo of [Hofstadter's law](https://en.wikipedia.org/wiki/Hofstadter%27s_law) and being lazy.
Even with those objectives I always procrastinated and I may have needed to use LLMs to do some work as a result.
Even with those objectives I always procrastinated and ended up needing to use LLMs to complete some work on time as a result.
This situation was not helped by me getting sick on the last week but thankfully the friend bailed us out.
## Tour of the code
## Lessons learned
[Link to directory with code](https://github.com/bucs445spring2025/portfolio-team4/tree/main/src)
In short the code is a mess.
### `src/client/*`
I don't understand anything here, I didn't write anything here and had little involvement in it's creation so I don't have any comments on it.
### `src/data/azure_function_timer/*`
This is code I wrote when I naively intended to do so much more with this project.
Specifically it's most of the work needed to poll the university buses data from a live map's API.
I was planning on gathering the data to do some machine learning to predict bus arrival times, this never ended up happening.
### `src/reverse-proxy/*`
Here lives the files for a caddy docker container which reverse proxies
### `src/server/*`
Here lives all the files for our API backend which... put simply I didn't work on this fast enough resulting in this code doing nothing useful to the end project.
#### `src/server/data/*` and `src/server/usable_gtfs/*`
Here lives static source data from the university bus system and the local county's bus system and my various attempts to deal with them.
GTFS is a file format google specified to allow transit angencies to describe how their transit systems work in a computer readable way.
If your local municipality has a transit system and you want to work with it via a computer program check to see if they just publish GTFS data for free with a quick search.
#### `src/server/stops/*` and `src/server/trips/*`
This... is bad.
My idea with these files is that they'd just be something I generate once from the aforementioned static data commit to source control and then never touch again, able to serve static files over the network for most of the API (designing the API appropriately).
That was foolish, the only things that should've been checked in were the source data and the code which turned that into the desired format with either a build or a startup step being used to generate these files.
This decision caused headaches like "$base_domain" and being really reluctant to ever make changes to formats for the api.
Hopefully this is a mistake I will only make with one project.
I don't remember why `src/server/GET_STOPS.json` and `src/server/GET_routes.json` exist but I think they are the same idea with the same problem.
#### `src/server/Dockerfile`, `src/deno.json`
Files for tools used for development.
#### `src/server/TODO.md`
reading this file I can only laugh.
#### `src/server/graph.ts` and `src/server/cypher-shell`
### Misc trouble
The friend uses a Windows laptop the class requires Docker (and I prefer it) this combination led to some annoyances for both of us.
Example: Annoying code
```ts
const usingDocker = !Deno.env.has("DISABLE_DOCKER");
const root_url: string = usingDocker
? Deno.env.get("ROOT_URL") as string
: "http://localhost:8080/api";
// Vs if this was docker only (probably could've simplified the above but oh well)
const root_url = Deno.env.get("ROOT_URL") as string;
```
But whatever it worked and wasn't make or break.
1) If I care about the success of something and nobody has taken charge of that thing within a timely manner, oftentimes this means within a week, then I need to take charge
2) Barring circumstances where it isn't possible to do so a MVP of a project should be completed as the first unit of work, in this case that would've been a week.
Afterwards units of work should be visible features.
3) Any problem which is not fixed will remain a problem.
In this case a team member having low remote responsiveness will not magically change so it'd have been better to start scheduling in person meetings sooner.
## What now
So now that the semester is over what will happen?
This work is probably going to be abandoned due to me needing to focus on other things.
If I come back to this project though I'm probably going to rewrite again due to probably doing it solo and wanting to yeet old mistakes.
If I come back to this project though I'm probably going to rewrite again due to probably doing it solo, choosing different technologies for front, getting rid of compromises which were made to accomodate development being done over Windows and Linux rather than just on Linux and backend and disliking some of the current solutions to problems.
Note to self: never check generated files into the build system, the disk/network cost(s) of git are not the reason they're just salt in the wound.
The actual reason is because generated files by nature of being generated are annoying to change so it'd be better to check the generator script in and have generating them as part of either process startup or a build step to put the friction where it causes the least damage.