Fork me on GitHub
#cljs-dev
<
2018-01-16
>
rauh07:01:35

@dnolen CLJS-2472 won't do a lot to the average CLJS user unless we also implement: 1. Custom Iterator for ChunkedCons/ChunkedSeq (probably easy). 2. A change to SeqIter that "escapes" to a fast iterator when the next is IIterable (same approach as my new seq-reduce that escapes to chunked sequences). I have a prototype for 2:

(consume-iter (iter (cons 0 (range 3000)))) ;; 300ms, NEVER escapes first/next due to Cons
(consume-iter (my-new-smart-seq-iter (cons 0 (range 3000))) ;; 24ms, escapes after one next call.
So over 10x faster. Impl is very easy.

rauh07:01:16

Q: Do you want 2 new tickets for this?

lxsameer11:01:51

it's a little bit old though

rauh12:01:17

Patch lgtm

dnolen14:01:20

@lxsameer I left some feedback

mfikes16:01:49

ClojureScript requires Java 8 now, right? (Owing to a requirement by Closure Compiler.) The first paragraph of Quick Start might be out-of-date with respect to this in indicating that only Java 7 is required. (Or, perhaps Java 8 is only required if you actually end up using Closure via some code path?)

juhoteperi16:01:35

Closure classes are imported by cljs.closure always, so I'd think Java 8 is required always

dnolen16:01:39

@mfikes I think it’s safe to say Java 8 is a hard requirement now

mfikes16:01:55

I’ll submit a small PR to update the Quick Start paragraph.

richiardiandrea22:01:03

Have you folks ever noticed the following errors:

Compiling out/cljs/tools/reader/edn.cljs
WARNING: cljs.core$macros/bit-or, all arguments must be numbers, got [string number] instead. at line 117 
WARNING: cljs.core$macros/bit-or, all arguments must be numbers, got [string number] instead. at line 117 
WARNING: cljs.core$macros/bit-or, all arguments must be numbers, got [string number] instead. at line 118 
WARNING: cljs.core$macros/bit-or, all arguments must be numbers, got [string number] instead. at line 118

richiardiandrea22:01:55

scratch the above, it does not come from ClojureScript code base

dnolen22:01:00

haven’t seen that before

dnolen22:01:27

well it’s tools reader so maybe we changed something

dnolen22:01:53

though we use it ourselves, so I’m surprised it hasn’t come up

richiardiandrea22:01:13

ok cool I will try to investigate more, thanks David

richiardiandrea23:01:22

It does not seem to affect ClojureScript (I bumped tools.reader locally) reported here: https://github.com/anmonteiro/lumo/issues/341