I'm a rails dev and I love the framework but there's a realist in me that sees something akin to the phrase "markets can stay irrational for longer than you can stay solvent"
I'm a rails dev and I love the framework but there's a realist in me that sees something akin to the phrase "markets can stay irrational for longer than you can stay solvent"
EDIT: typo
Hotwire is the polar opposite of spaghetti code
Following variable life cycles in Ruby in general is much harder than a typed language. There are tools like pry for this sort of thing, but that’s more work than just having types with good ide integration.
It’s underrated since devs mostly do it without thinking about it, but in enough cases you need to double-check the type before running any operation that requires a certain type. Unchecked it will either crash entirely or cause bugs so esoteric that they are very difficult to debug.
I don’t think I’d categorize it as “how Rails suffers without types”, more like “stay away from the edges and you’ll be fine”, but it can still bite you hard if someone mis-steps.
Anything I can think of will immediately throw an error or would be covered by the most basic of test suites
That’s exactly what we don’t want in production code. We want it to work without errors.
An example: assign a variable to the output of a library function, then pass the variable in to a different function. Super basic, super common.
One instance of that is calling the first function and expecting an array back but getting any one of ‘String’ (edge case that the library maintainer decided is a good idea) or ‘False’ (maintainer decided that’s what happens when something fails during the call).
If you call ‘.each’ within the second function without verifying that the variable is iterable first the function will throw. (I’ve even seen a couple rare cases where a custom class responds to ‘.each’ but is not properly iterable).
Most Ruby devs will “just” check before they call ‘.each’, or wrap the second call in a rescue block, but that’s exactly the sort of potential for misstep I’m talking about.
I think the current mania for types has arisen largely out of the js SPA crowd creating problems for themselves doing the opposite,
Simple technologies leave innovation points to be spent in the problem and user instead of experimenting with new tech that is usually a rewrite of something else that’s already similar, except without the hardening and maturation
In a one person team merging your server with your client and using HTMX can be simpler (depending on if the value is in your UX or in your logic/data).
However as the team gets larger then I find having a clear separation between your API(s) and your Client(s) is much more simple and maintainable. Also it can unblock product decisions like offline mode or PWA's.
Finally its worth stating that being in an SPA community like React can give you access to "economies of scale" in terms of the libraries available to save you time (Tables, animations, UI kits, Video players, etc.) allowing you to do more with less people.
The problem is that the SPA way of thinking has infiltrated every nook and cranny in this industry. The ratio of APIs vs clients consuming those APIs come scarily close to 1, at least with the customers I work with.
Now you have to maintain two projects, two CI/CD pipelines, two hosting solutions, potentially two sets of languages and tooling…