Clojurescripters.... feedback on these two questions?

* Q: Which ClojureScript tools do you use for interactive development?
** ClojureScript's built-in REPLs
** Figwheel Classic
** Figwheel Main
** lein-cljsbuild's command-line REPL
** shadow-cljs
** Dirac
** Devcards
** cljs-devtools
** Other [open response]

* Q: How are you running your ClojureScript tests?
** Headless browser (or actual browser)
** Node.js
** Nashorn
** Devcards
** Phantom
** doo
** Karma
** shadow-cljs
** Kaocha
** nubank/workspaces
** Figwheel
** I don't have any tests!
** Other [no response]

the first is in replacement for the prior "CLJS REPL" question, both have updated answer sets. what's missing?

@alexmiller cljs-devtools is a library that is usually used during development with any other tool. doesn't seem to fit into the question?

also the first answer should probably be cljs.main (as that covers the REPLs and more)

devcards also seems to fit better into the second category, not the first

I’m curious (no idea if anyone else is or if it’s relevant) how many people are coding or would like to code more cljc based code. I personally aim to do most of my development in that space and I think knowing whether it’s a niche area or not would be useful? Not sure what’s currently in the survey, so no idea if it’s already addressed…

or combined variant:

* Q: Which ClojureScript tools do you use for interactive development and testing?
** cljs.main
** ClojureScript's built-in REPLs
** Headless browser (or actual browser)
** Node.js
** Nashorn
** Phantom
** Figwheel Classic
** Figwheel Main
** lein-cljsbuild's command-line REPL
** shadow-cljs
** doo
** Karma
** Kaocha
** Dirac
** Devcards
** nubank/workspaces
** cljs-devtools
** Other [open response]

(multi-select question)

> Q: Which ClojureScript tools/libraries do you use for interactive development and testing?

dunno if it makes sense to separate but in my view some of these don't qualify as "tools", rather something you use with those tools.

kind of taking a big tent approach here

yeah, just meant the wording of the question

its fine really ... nevermind me 😛

both of the original questions (about CLJS repls and CLJS testing) were specifically added to assess the status quo wrt how people were doing these things (and pre-dated things like Figwheel)

yeah there is lots of overlap in the answers so I guess its kinda difficult to have "generic" answers

eg. when using karma you are likely using a headless browser

or when using kaocha I think you that wraps doo ... which wraps a headless browser

but its good to list stuff that people may also be using separately with handmade tools

I'm looking at changing them because I no longer felt like the REPL question was actually reflecting how people work. yes, they use REPLs, but in the context of larger environments or in combination with other things (and maybe not even just on ekind of RPEL). Maybe similarly for testing where something like Devcards seems to be equally as useful, but very different in some ways than traditional "test runners".

so my meta question is: what questions are useful to ask to assess how people work with developing ClojureScript apps?

Part of this is also tooling I think, I didn’t realise the value of a clojure repl vs say a python interpreter until I could get inline send form to repl working, it’s still really not great in clojurescript in my experience… IE, people with different workflows will have wildly varying perspectives…

I would like the results of this question or questions to answer "how do people usually work with ClojureScript?"

are there families of toolsets that would make sense to combine?

assuming you already have a separate question regarding build tools it might make sense to not repeat them in that question

and instead focus on the feature they provide, eg. "Hot-Reload" provided by the build tool (figwheel or shadow-cljs)

same for REPL

hot-reload is a big part of my workflow ... I barely use the REPL for CLJS yet the REPL is my primary tool for CLJ

npm usage versus cljsjs?

for accessing js libs?

yeah that would be interesting too. I would be curious to see how many actually use the recommended webpack+cljs.main combination vs cljs.main+cljsjs

Is webpack+cljs.main recommended? I don’t think I’ve ever seen this…

is this a dep mgmt question?

where are the deps defined in any of these cases?

package.json with npm or yarn. when using shadow-cljs or cljs.main + webpack

so as additional options to "What tools do you use to define and download your dependencies?"

** npm (package.json)
** shadow-cljs and yarn
** webpack and cljs.main

(other options are more java focused - clj, lein, boot, maven, etc)

sorry. JS deps are always managed by either npm or yarn (which both define the dependencies in package.json)

you typically use one of those and then consume the packages with either shadow-cljs OR cljs.main+webpack

can you help me rewrite those answers?

define/download is only npm or yarn

I guess for the purposes of that question, probably only npm or yarn (package.json)

yes

I have another question "What tools do you use to create project artifacts for release/deployment?" and I will have both shadow-cljs and cljs.main in there

do you have shadow-cljs in the "define and download dependencies"? it can do that as well 🙂

where are they defined in that case?

shadow-cljs.edn :dependencies. same format as lein :dependencies (same library in fact, pomegranate)

lots of people use shadow-cljs either standalone or with deps.edn/project.clj. would be interesting for me to see some numbers behind that 🙂

@alexmiller we use cypress as well for testing UI flows in the browser - i’m not sure if that’s helpful for this survey though

i guess that’d be in the open response bit

a few people mentioned that in last year's open responses but not enough that I had added it

by "a few" I mean like 3

np, it’s a generic browser testing tool

i’d love to know more about how people handle errors / failed computations

nil punning? bail on false? exceptions? failjure?

nested ifs? 😄

sorry, not going to expand to things like that

fair enough.

nil punning and exceptions are both idiomatic in clj / cljs

condition systems and monadic failures are not idiomatic, but feasible and some people do those

i guess it’s another discussion

interesting one though

doing it cross platform too..

anyway

phantom falls under the category of Headless browser (or actual browser) , doesn't it?