00:00 <FromGitter> <mgarciaisaia> Anyone here? We're live at hangouts for the Crystal Q&A - join us at https://www.youtube.com/watch?v=E25AGpYyQw0

00:59 <FromGitter> <bew> About codegen, iirc the compiler generate *a lot of* duplicate code: for a method that that is used by multiple types, the generated method is copy/pasted for each types using it... will this change?

01:10 <FromGitter> <megatux> even with --release flag? :/

01:13 <FromGitter> <bew> I don't know, I've read that somewhere.. But iirc, the `--release` flag only remove debug from generated LLVM IR (and maybe other things too), and enable optimization during LLVM compilation.. But the defs codegen method doesn't change imo

02:45 <FromGitter> <drosehn> Thanks for the live Crystal Q&A. Unfortunately for me I was too busy to follow it live, but I did watch it a short time after it was finished.

09:08 <FromGitter> <sdogruyol> hey everyone

09:14 <RX14> hi @sdogruyol

09:15 <FromGitter> <sdogruyol> hey @RX14 how you doing

09:22 <FromGitter> <bararchy> Hi Guys, how did the 3AM Q&A go ? XD

09:22 <FromGitter> <sdogruyol> it's good

09:28 <FromGitter> <emilebosch_twitter> Hey all, any dutch crystal fans wanting to meet up?

10:04 <FromGitter> <RX14> @sdogruyol I'm doing fine! I didn't try and stay up for the Q&A this time :) I was about half way through then I had to go out.

10:17 <FromGitter> <sdogruyol> We talked about you

10:19 <FromGitter> <RX14> Yeah I heard.

10:21 <FromGitter> <RX14> I watched the first half of it

10:21 <FromGitter> <RX14> This morning that is

10:22 <FromGitter> <sdogruyol> Cool

12:31 <FromGitter> <bararchy> I'll watch it later today , any talk regarding concurrency ?

14:00 <FromGitter> <straight-shoota> Is there an easy way around this problem? ⏎ ⏎ ```code paste, see link``` [https://gitter.im/crystal-lang/crystal?at=59301de05e34568d5eb1c2f6]

14:00 <FromGitter> <bew> hmm maybe declare it before the measure block as nil, then use `not_nil!` after when accessing it?

14:01 <FromGitter> <bcardiff> `foo = uninitialized Foo` before the block

14:02 <FromGitter> <bew> :+1:

14:02 <FromGitter> <straight-shoota> Wouldn't it be easier if `Benchmark.measure` was a macro?

14:02 <FromGitter> <straight-shoota> But thanks, that uninitialized works

14:19 <FromGitter> <straight-shoota> Still thinking about it... any reason for `Benchmark.measure` *not* to be a macro?

14:20 <FromGitter> <mverzilli> why would it?

14:28 <FromGitter> <straight-shoota> Because then the block would be expanded in the current lexical scope and you wouldn't need `uninitialized` to assign values inside the measure block: https://carc.in/#/r/249m

14:31 <FromGitter> <straight-shoota> Maybe it's a bit far fetched and not worth the additional effort for macro expansion.

14:33 <FromGitter> <mverzilli> yes, that last thing is why I asked... to my eyes using variables outside their lexical scopes looks a bit ugly

14:34 <FromGitter> <mverzilli> plus we generally try to avoid using macros unless they provide a lot of value... in this case we would be using a macro to sort of "break" lexical scopes

14:37 <FromGitter> <straight-shoota> This looks even better, without the yield there isn't even a need to initialize the variable beforehand: https://carc.in/#/r/249x

14:38 <FromGitter> <straight-shoota> But your probably right...

14:39 <FromGitter> <straight-shoota> The effort for macro expansion seems not worth to put it in stdlib

14:44 <FromGitter> <straight-shoota> Returning a tuple would not require macros: `time, foo = MyBenchmark.measure { "hello" }`...

21:36 <FromGitter> <elorest> Is there an existing way to get the time that a HTTP::Request is initialized for more accurate logging?

21:37 <FromGitter> <elorest> Everyone including kemal seems to start the time in the first handler but that’s actually skipping a decent amount of data processing time.

21:38 <FromGitter> <sdogruyol> @elorest how would you access context outside of handlers?

21:52 <FromGitter> <eliasjpr> The best way that I can think of getting accurate http request/response lifecycle would be to initialize the server with a block and timing the content of the block

22:21 <FromGitter> <elorest> @sdogruyol I’m not saying that you should access the context ourside of handlers, but the time starts well before context gets passed to the first handler. this leads to drastically different times when benchmarking vs log output.

22:22 <FromGitter> <elorest> @sdogruyol The idea that’s stuck in my head is saving the time that HTTP::Request is initialized on the request and then comparing it in the log handler.

22:25 <FromGitter> <sdogruyol> Makes sense but IDK if it's worth it

22:28 <FromGitter> <elorest> Probably not. If i have nothing better to do some day I’ll probably take a stab at it.

22:28 <FromGitter> <elorest> I’m looking through the rails loggers now

22:54 <FromGitter> <carsonbaker> ```code paste, see link``` [https://gitter.im/crystal-lang/crystal?at=59309b1feb257a8725ccbf12]

22:54 <FromGitter> <carsonbaker> is there a better/easier/cleaner way to read in this file as a Slice(Int16)?