Fork me on GitHub

@jlmr after some debugging, I found out that datahike.db.DB re-writes seq so it returns something that's not the "key-value" as UNREPL expects. I'm seeing if there's any way to fix this for now 🙂


@mauricio.szabo awesome. Does it rewrite “a seq” or the entire seq protocol?


I'm really thinking of a workaround: even thinking that this could change in future implementations on datahike side, I can't have an "unparseable EDN" breaking the entire plug-in....


Agreed, even a message saying something like “Couldn’t parse edn” but just continuing to function would be nice.


In Chlorine, advantages/disadvantages of “Should we refresh namespaces when we save a file?”


The advantage is that you can be sure that when you save a file, Chlorine will try to refresh that namespace using (require '[<namespace> :reload :all]) or


The disadvantage is that it's kinda fragile: you have to be sure that when you save a file, you don't break anything. Otherwise, not even autocomplete will work 😞


When I started to work with Clojure I always left it on. Nowadays, I left it off all the time, because when you make a mistake its really hard to fix when you leave it on...


BTW, autocomplete will not work if you're using the "" approach, because it destroys namespaces and re-loads then 🙂


Yeah, using simple. Been playing with turning off the refresh


I am fairly vocal against all of these auto-refresh/reloaded approaches. It can sometimes make things seem "magically" easy but when it goes wrong it can really confuse people -- and it is certainly not simple (in the Clojure sense).


Stu Halloway kind of speaks against these approaches too in his REPL talks, and Eric Normand isn't in favor of them either (re: his REPL course online).


ProtoREPL also had similar options that Jason Gilman's "opinionated" guide said should be "on" but which caused users all sorts of weird problems (and, again, I advised ProtoREPL users to turn them "off"!).


Yes, even while working on ClojureScript, sometimes if there's an error on the code, it'll not reload in the browser / node.js environments. This means that ClojureScript REPL will not work (and sometimes, autocomplete/etc will also fail).


One thing I’m wrangling with is tabs/spacing. I use parinfer with experimental on (nice!) and paredit so I can slurp, barf, etc (also nice!) but I often see weird tabbing and indenting. Often it will indent 1 space, even though set up for two. Tabbing usually goes 2 spaces, but sometimes goes 0 spaces (!) then jumps four. Sometimes one. I have no idea what is going on. I expect this isn’t Chlorine, but asking here. Just redirect me to another forum if needed


It isn't Chlorine, but no worries. Clojure indentation sometimes use only one space, other two. It depends: there's some configuration on lisp-paredit where you can change which forms will indent 2 spaces


Also, Atom sometimes does strange things too, unfortunately 😞


Also, I think Parinfer and Paredit can compete/conflict sometimes, which doesn't help. Parinfer's "smart" mode isn't always very smart. I often find myself selecting a block of code and using smart reindent explicitly to fix this -- and I've had Parinfer break code by adjusting parens to match the (incorrect) indentation. It can be quite frustrating. But, overall, its benefits outweigh its downsides so I still keep using it... 😐


Yes, for example, I'm used to use CTRL+ALT+9 to change parinfer mode from "smart" to "paren" before pasting any code. This wil not change the structure of pasted code, so I can indent manually after and then change back to "smart"


I forget to do that a lot of the time. I probably ought to set up a command that pastes code with Parinfer mode set to paren (and then changes it back) -- and get used to whatever hot key I bind to that.


I was thinking about a plug-in for Atom that would evaluate multiple commands, one after another, exactly because of that 😄. But in the end, for the cases that I had to use it, I ended up using a custom command on