With summer right around the corner I'm hoping to spend at least a little time on some personal coding projects. There are a few work related tools I'd love to develop and just some random areas of CS I'd like to explore. If I finish them, the work projects will be web based.
I was thinking about using this as an opportunity to do a deeper dive into Clojure having used it for some experiments and competitions like Advent of Code but at the end of the day I decided not to. It's not that Clojure wouldn't be a good language for web apps. It looks like it would be great. Rather, I'm finding the ecosystem and build systems in particular to be more trouble than they're worth for a side project - particularly when I want to accomplish more than becoming comfortable with the Clojure ecosystem.
Things were so much simpler in the old days. Way back when I was a software engineer at Goldman Sachs I worked on a couple of large, complex systems. All I had were a bunch of C files in a bunch of directories and a Makefile to build everything. Simple and straightforward.
If I'm working in C or C++ I can still do the same.
In Java I was always able to do the same since Make seemed to be accepted in the community. Now there seems to be many more options - Ant, Maven, and Gradle to name just three.
Python was even easier - just set up your directory structure
__init__.py files and your good to go and by using a virtual
environment, which is pretty easy, you can keep the entire development
Nowadays things are much more complex. Sure, I can create a simple Clojure program and run it but anything bigger and you need a build system. For a long time the tool was Leiningen. You define a project file using it's particular syntax and it can download dependencies, manage builds etc.. It actually doesn't look so bad if you're living in Clojure all the time but for a hobbyist like me I found myself looking for other peoples templates and hoping for the best. Whenever I needed my project to do things outside the norm I was pretty much out of luck.
Then came Boot as an alternative. Where Leiningen used a descriptive configuration file, Boot used a configuration file that was basically Clojure code. It was different but to me, it had the same problems - documentation was lacking and too much magic going on behind the scenes. Finally, Clojure started to come with build system. I want to like it but the documentation is even sparser and with three competing build systems with varying levels of use and documentation it's less than ideal for a guy like me.
I don't mean to be unfair to Clojure and its ecosystem. I still really like the language and if I lived in it I'm sure I'd figure it all out and be happy with it. It's also no worse and arguably better than ecosystems for other languages.
There's really no easy answer.
Putting my teacher hat on for a second this is also an issue that our students will have to deal with at some point. Are we helping them by keeping them in web based language silos and simple IDEs or hurting them? At what point do we transition them or do we leave it for them to do on their own? I'm not taking any position on this here and now but it's something worth thinking about.
Back to my side projects.
So, for the sake of getting things done I think I'm going to fall back to Python. I don't need to deal with any complex build systems, I can write my web applications with Flask and it's also a nice platform for the other things I'd like to do.
I'll still probably play with Clojurescript for the web frontend ut if that turns out to be proiblematics I'll just go back to VueJS.
Time to get to work.Tweet