Hello everyone, first post over here, I’m a junior dev that was working on an open source picture organizer based on location groups within sub-directories, like streets and cities (inferred from EXIF geotags being extracted and sent to a reverse geocoding API) which I’ve first coined as GeoPicSorter (outlined in its home page). There’s a TL;DR at the bottom if you want to get right into my questions, but I felt like talking about my tool could be of help and worth it in case anyone ever wants to help with its development, anyway…
To develop it, I used Java and NetBeans (as the only IDE I found with a free GUI editor at the time I started ― at least over a year ago) and GeoPicSorter accomplishes the goal I needed it for, however, as I updated it, I started to feel like I’ve underestimated how hard to maintain and messy code becomes for the GUI part. NetBeans had its own gimmicks like making it read-only, fully naming the Swing classes every single time they were called, etc. The Swing GUI also feels kind of uncomfortable to me (specially in the folder selection part where I choose a folder with the pictures to organize, it’s too compact/constrained) and I’d like for my program to use the system GUI, or at least something that makes it feel classic yet user-friendly. But it’s not like Java is the worst option to use, I just feel like there could be better options suited for the project, and I want to learn a new language in the process.
I’ve attempted to re-write my tool (and rebrand it in hopes the name can be shortened without conflict with “GPS”), to write cleaner code while attempting to retain cross-platform compatibility (which ideally, I’d like my tool to have so more people can try it. I’d like it to work at least for Windows, and hopefully Linux without Wine, maybe Mac too). It’s not a necessity though, just would be great to reach more users. The re-write would be available at a Codeberg repo.
For the re-write: I’ve tried Dart, Python, and C++ so far, however, I struggle with different things in each language that make it impossible for me to figure out on my own with my experience, and I feel like I’m wasting time going in circles. The problems I have are:
- Dart: I haven’t touched it in a long time, but I believe my issue was the lack of documentation and functions/methods for the only EXIF reader I could find, and how it extracts the coordinates (into an array-like set of 3 numbers that represents latitude/longitude in DMS format) without functions to handle them and convert them to DD easily.
- Python: while it was a satisfying language to learn and experiment with (and had everything I needed), I couldn’t figure out how to make an executable that could detect (and depend) on the system installation of Python (instead of bundling one), I’ve exhausted every EXE-generating option I could find, and none worked the way I needed it to. If there’s one you think I possibly didn’t try, I’d love to know!
- C++: I’m having trouble to start a project properly. I’m pretty much a newbie with CMake (which seems I need to use if I want to make a cross-platform C++ project), and my greatest struggle was setting up the dependencies so as to not to give a hard time to new contributors to join or make one-time contributions (and when I move to another setup). I’ve tried Conan, and its repository has everything I needed (including Qt for the GUI), but for some reason it’s been a struggle to get the dependencies (and the project itself) set up around it. I’m using Visual Studio for now since I found it best to get into C++ and it was recommended, but I’m open to other options. Qt always failed to install via Conan on the command line (IIRC, the error was just a generic one and occurred after waiting for the download and install).
So, I’ve been using FLTK, but the GUI options and components feel very limited, and I’m stuck with VS2017 and Conan 1.x, because the add-on for VS is outdated, and I couldn’t integrate 2.x into VS.
After all this context (which I’m sorry it dragged for so long already), I can finally ask: what can I do to properly start my project from scratch? Or is there a language/framework that’ll best suit my needs? The dependencies it has are already mentioned on the home page, but the EXIF reader should be able to at least make it possible to get the coordinates in DD format, and consider the reference cardinals for latitude and longitude before the conversion from DMS format.
Should note that I’m determined to see my project through, I’m just lost on what to do and in need of advice.
Any help is appreciated, thanks in advance!
TL;DR: I’d like directions on how to set up a software project with dependencies, so that future contributors can pitch in with the least amount of set-up required when they clone the repo. The original program is an open source location-based picture organizer for desktop systems, and I want to rewrite it with a different set of tools (preferably not Java with NetBeans), its code will be hosted at Codeberg.
First off, cause you are programming under windows, a lot of things will be harder for you. As seen on your problems with Python.
Most Linux installs have it right from the get-go and everything else is as simple. So giving directions for developers on other platforms might be much easier than what you had to go through. (Maybe use WSL?)
Let’s get to your real question: How does one organize dependencies in a way easy for new contributors?
Since you will use Python, I will use that as example.
Most languages have a way to automagically import dependencies. Python has the requirements.txt file. Installing dependencies is then really easy. It is also a widely known way to do that, has lots of explanation online etc. so seasoned pythoneers will know what to do and younglings will get to know a good standard right away.
Bonus tip: If you don’t have a GUI library yet, maybe also search for game engines. They provide all the necessary tools as well, oftentimes have good GUI add-ins and are (mostly) for all mayor platforms.
Thanks. I understand, with your example and using WSL, I might have some luck with Python (I’ll have it into consideration), I still have the code for that re-write attempt and it was going pretty well. Hopefully I’ll figure out the packaging part if I go with Python because I want to version my program.
For the GUI, I had the built-in Tk libraries, but I appreciate the tip neverthless (will be useful if I end up going with another language). Reminds me I was considering to use ImGui for C++, but ended up using FLTK.