No, it’s still not full agreement. While I agree there are apps that never need scale, we, as a software engineering community, are horrible at predicting this. I don’t trust any of us, not even myself. For something that we don’t think will need to scale, let’s add a new controller to an existing Phoenix app. However, if the company has no existing Elixir in production, my opinion is that it’s a bad idea to introduce a small endpoint that’ll never scale to their portfolio of technologies. As much as I love Elixir and want to see it grow, it does not leave a good impression as “that one app the passionate engineer talked us into trying, then left us with no one else to maintain it”.
gregvaughn
dimitarvp
Hm, I am not sure why you brought different programming languages / frameworks into this but I don’t disagree with what you said there either. 
Sure. Same here. But I will gradually, over time, be chasing after using technology that spares my brain. Currently the amount of files I have to follow in order to be truly productive in Phoenix is unacceptable for me. Nobody has to agree but I’ve experimented all over the spectrum and “less files” is my best productivity enabler so far.
And as mentioned a few times, I don’t work with Phoenix’s specifics every day, I am a backender / data guy / CLI tools encyclopaedia and an alright Linux sysadmin – and I’ve never been impressed by almost any web framework; they leak too much of HTTP’s idiosyncrasies into what should be strictly business domain code, and Phoenix is no exception.
But I think we as senior devs understand that leaky abstractions are one of the most difficult problems to solve. Hence I don’t complain too much and do my best to keep discussions on the topic productive because I know this stuff is mega-hard.
But I’m free to pursue ways of doing stuff that make me smile, be excited about programming again, and also be productive – because as much as I love flinging characters on the screen, I also want to make money and not be a hobby computer nerd until my grave. There are many more things to life than this that also make it a joy to live through, and trying to be more productive is one way to help us pursue those instead (== less hours in front of a screen), and not the ever-elusive 100% mastery.
2
dimitarvp
@taro Sorry for off-topic, got carried away.
@AstonJ feel free to separate comments in an off-topic thread if you feel it must be done. I simply enjoy chatting with @gregvaughn so much. Or if not, feel free to delete this comment because it is definitely off-topic.
1
gregvaughn
You absolutely are. However …
This does not resonate with me. I almost never notice the organization of source code on the file system any more. I don’t have anything to offer to help. So, that’s why I’m bowing out of the conversation.
2
dimitarvp
I use Live Grep in my IDE as well.
But I want to be able to fit a small app into one screen. For me it’s a hard requirement.
Thanks for engaging. 
3
taro
Oh I appreciate every single comment. I read and learn and take notes from all of them. Thanks for sharing your wisdom. I hope moderators are happy too. Your off-topics are inspiring. I took a note of DSL for text protocols.
Please speak up. Difference is the very reason we should be talking. It’s probably only thing I’m bringing to this community. No one has to get dissuaded. I appreciate that you have chimed in when it’s rainy. I have my hopes for Elixir adoption. But I can’t do it alone. So I started the discussions in this community.
I started web development learning SvelteJS. SvelteKit is amazing in its way but there was not enough batteries included for my project. I wanted to learn a major framework and an awesomer language. Found Phoenix on Fly.io and Elixir was perfect. I understand the benefits and wisdom behind full-scale frameworks just that much.
As I come from more design disciplines, I find some software development practices are destined to overload one’s cognitive capacity. They would diverge into infinity without coming back to a simple principle. When it hits the critical point, it becomes an abomination that no one understands. And I see the effort to mitigate this in, for example Saša’s Boundary and his posts.
The source of this complexity doesn’t seem trivial and I haven’t fully explored this. But I wonder if it’s rooted to the file system of OS? Files are often grouped by their categories in a project, instead of their context. spawn feels weird to me because among all the functions that describe logics and data flow, it describes processes, which feels like another dimension of the system. Yet they are all written and managed the same way in the files of the directory system.
4
jam
I like this. It’d be cool if you extended this idea for a channel router as well, maybe borrowing ideas from ChannelHandler with some adjustments on the syntax to fit Mar.
3
jam
Maybe it could use temple for an all-elixir feel. Something like this:
defmodule MyApp.Post do
use Mar, path: "/post/:id"
# using the data-star example on the home page
get "/endpoint" do
sse.patch_elements(div id: "hal" do: "I’m sorry, Dave. I’m afraid I can’t do that") # maybe get rid of sse here and just call it patch_elements. would need to automatically wrap anything in here with `temple do end` I think
:timer.sleep(1000)
sse.patch_elements(div id: "hal" do: "Waiting for an order...")
end
temple do
h1 do: "Post for ID #{id}"
div id: "hal" do: "Waiting for an order..."
button data_on_click: "@get('/endpoint')" do: "Open the pod bay doors, HAL."
end
end
1
taro
Exactly! I’m using bare Plug, HTMX, and Temple now.
I’m not actively working on this topic. But I will when the time is right. Thanks!
edit: Datastar sounds like an upgrade from HTMX. Cool stuff, thanks!
2