<FromGitter>
<Blacksmoke16> that doesnt seem right
<FromGitter>
<Blacksmoke16> given larvel is 2nd to last maybe just a bad run?
<FromGitter>
<s0kil> "Results are not production-ready yet" ⏎ I have seen the results drastically change over time as they adopt different benchmarking techniques.
<FromGitter>
<Blacksmoke16> seems to consistently in the bottom 5 tho
<FromGitter>
<Blacksmoke16> like less than 2k
<FromGitter>
<Blacksmoke16> worth some investigation to determine if its something with the benchmarking that could be improved, or a identify something that could be optimized
<FromGitter>
<s0kil> Lucky is the last one, strange,
<FromGitter>
<Blacksmoke16> yea, i would like to think it can do more than ~300/sec ha
<oprypin>
sgt_chuckles: just use CrSFML what's the problem
gangstacat has quit [Ping timeout: 272 seconds]
gangstacat has joined #crystal-lang
<jhass>
s0kil: Blacksmoke16: the big issue with type inference for docs is that Crystal works by only instantiating called methods. What should it do for a method like def x(a); a.something; end;? The possible return types depend on what kind of types are passed and that's only known once you call the method with things and changes as you call it with different things
<raz>
but doesn't the compiler have the knowledge anyway?
<jhass>
I just explained why it does not
<raz>
then how does it know and complain when i write that method and `a` doesn't have a something method?
<jhass>
it does not until you actually invoke the method
<jhass>
here it even could know that String doesn't have yeah_nope
<raz>
yup, but for that case the docs could contain the types, right?
<jhass>
but it never even checks until you call
<jhass>
what types?
<raz>
the argument and return types :)
<jhass>
what would they be?
<raz>
the ones that the method was called with and returned!
<jhass>
but in a doc context most methods are never called
<raz>
maybe docs are generated in the wrong context then ¯\_(ツ)_/¯
<jhass>
and even if they would, it doesn't mean they're called with all possible or even realistic types
<jhass>
what would be the right context?
<raz>
that's true, but it would still be more convenient than having to manually annotate them
<raz>
perhaps spec context
<jhass>
this is not an argument of convenience but about possibility
<raz>
that would also encourage people to improve their spec coverage
<jhass>
and still it would be a heuristic and lead to misleading/wrong docs in some cases
<jhass>
"Oh I cannot pass an X? Bummer" in fact you can, the specs just never idd
<raz>
true, these rare cases could then be manually annotated
<raz>
in the general case my specs exercise the methods in all ways they are intended to be used
<jhass>
I disagree this is an awareness we can reasonable spread as the community grows
<raz>
i dunno, just throwing ideas out. i would find it convenient not having to annotate all my methods to make the docs look right
<jhass>
wrong docs are 1) an subtle error 2) more harmful than no docs
<raz>
yup that part is true
<jhass>
I'd argue the docs can look perfectly fine without types :)
<raz>
they can, but i'm frequently annoyed when i can't click on the return type to see how to deal with it
<raz>
but i'll concede that's a minor thing in the grand scheme of things
oddp has joined #crystal-lang
<FromGitter>
<Daniel-Worrall> This Thread stuff on gitter confuses me
<jhass>
yeah just avoid it, it doesn't work well with the bridge anyhow :P
<FromGitter>
<Daniel-Worrall> I keep getting unread messages that won't clear cause they're part of threads
<jhass>
come to IRC, problem solved :D
<FromGitter>
<Daniel-Worrall> yeah, and more problems gained
<jhass>
problems like not being forced to a single UI or not having enough tabs and hungry webapps in them open? :D
<raz>
yes, come to the dark side! o/
<jhass>
or white or purple or whatever you want your client background to be! xD
<FromGitter>
<ImAHopelessDev_gitlab> hungry webapps? that's a nice euphemism for electron bloatware
<FromGitter>
<j8r> @Blacksmoke16 seems to be, thanks! Remains to figure out how to decode in JS
<FromGitter>
<j8r> I don't know what's the format
<FromGitter>
<j8r> I can post this in the forum, will be better
<FromGitter>
<Blacksmoke16> the format should be `BigEndian` or `LittleEndian`
<FromGitter>
<Blacksmoke16> prob a node module out there somewhere to do this?
HumanGeek has quit [Remote host closed the connection]
HumanG33k has quit [Remote host closed the connection]
<FromGitter>
<Blacksmoke16> > *<jhass>* problems like not being forced to a single UI or not having enough tabs and hungry webapps in them open? :D ⏎ ⏎ Just download more RAM 😉
HumanGeek has joined #crystal-lang
HumanG33k has joined #crystal-lang
chachasmooth has quit [Ping timeout: 272 seconds]
<FromGitter>
<j8r> Thats not really the format, but the endianess
<FromGitter>
<j8r> By format I meant how it is represented with bytes
<FromGitter>
<christopherzimmerman> I feel like I'm missing something obvious
<FromGitter>
<s0kil> > I'm right that the most efficient way to transmit information between crystal server and browser, websocket with Bytes protocol? ⏎ ⏎ On top of that `deflate-frame` WebSocket-Extension
<FromGitter>
<Blacksmoke16> not sure i follow what the goal is there @christopherzimmerman
<FromGitter>
<Blacksmoke16> `nt[{{key.symbolize}}]` would already return the value of the given key
<FromGitter>
<christopherzimmerman> `nt` is the current named tuple, I want to call a function on that value, based on what is provided to `fns`
<FromGitter>
<christopherzimmerman> So for the key `foo`, call `sum`, and for the key `bar`, call prod
<FromGitter>
<Blacksmoke16> so like ruby's `send` method?
<FromGitter>
<christopherzimmerman> I'm not too familiar with Ruby, but from what I can tell it's similar
<FromGitter>
<Blacksmoke16> think you would have to change something tho, `U` is a type, not a value
<FromGitter>
<Blacksmoke16> so you're iterating over `NamedTuple(foo: Symbol, bar: Symbol)`
<FromGitter>
<Blacksmoke16> so would have to actually get the value from `fns` argument, but then you cant do what you want because crystal doesnt have send
<FromGitter>
<eliasjpr> I have a private repo and shards install from within a docker container is not working as soon is public everything works. What I am doing wrong?
<FromGitter>
<christopherzimmerman> Yea, I don't think there is a way to call a function by name, maybe if I converted everything to a macro? But that would be pretty messy
<FromGitter>
<Blacksmoke16> not working how? says it doesnt exist?
<FromGitter>
<Blacksmoke16> does the docker user have access to it?