Fork me on GitHub
#cljs-dev
<
2017-09-14
>
lxsameer13:09:01

the goal is to remove duplicated externs, right ? I think cljs.closure/load-externs would be a great place to add this feature

lxsameer13:09:38

is that ok to break that function to smaller pieces ?

dnolen13:09:47

@lxsameer if that’s necessary - if all we need is distinct I wouldn’t bother

lxsameer13:09:07

good , thanks

anmonteiro17:09:37

@dnolen recently had some trouble with cljs.test/async

anmonteiro17:09:53

why does the body of the async macro need to be truly async to avoid stack overflows?

anmonteiro17:09:25

I thought it would just work for anything that accepted a callback. I was wrong

anmonteiro17:09:42

In my case all tests worked on Linux/ mac but failed on Windows

dnolen17:09:34

Async body must be async

dnolen17:09:42

If it's not don't use it

dnolen17:09:40

Callback will increase the stack

dnolen17:09:49

What else is possible?

anmonteiro17:09:20

sure, callback will increase the stack, I get that

anmonteiro17:09:46

just took me a long time to track this down, esp because Windows was the only outlier

dnolen17:09:45

Yeah I don't see anything obvious. Maybe possible to introduce trampoline but haven't thought about it

lxsameer18:09:43

guys what would be the next step for this ticket https://dev.clojure.org/jira/browse/CLJS-2166 ?

dnolen19:09:34

other than being patient

dnolen19:09:55

sometimes it takes a long time to get to stuff because priorities

dnolen19:09:03

in this case I’ll probably take a look tomorrow

favila19:09:53

@lxsameer the patch has some problems, I left you feedback

favila19:09:38

@lxsameer also why do you try to catch typeerror?

favila19:09:59

under what circumstances could instanceof throw TypeError?

dnolen19:09:52

I agree with those comments

lxsameer20:09:19

thanks guys

lxsameer20:09:53

@favila because other types for (e.g numbers) triggers the type error

lxsameer20:09:34

what's the desirable indentation size ?

lxsameer20:09:45

also regarding to clojurescript-version, what should be the number ? 909 ?

lxsameer21:09:27

but i think i shouldn't include that in the patch

favila21:09:39

your patch should only have added lines, no deleted lines @lxsameer

favila21:09:28

(def *clojurescript-version*) should never be committed, it's populated for built cljs.jar only

lxsameer21:09:43

@favila cool, thanks man

bhauman22:09:13

Ahh I see I'm too late, I just found the problem in util/topo-sort that caused it to be so inefficient https://dev.clojure.org/jira/browse/CLJS-1228 the same algorithm is being used in Figwheel and I finally got around to fixing it ...