08:12 <crystal-gh> [crystal] akiicat opened pull request #4389: Fix HTTP::Params.parse query string with two '=' (master...master) https://git.io/v9rzZ

08:31 <FromGitter> <crisward> For lower traffic sites, does anyone see any down side to building sites without the --release flag? Crystal seems to still be very quick without it, and I'm wondering if the added compile time is worth it. I'm actually considering using non-release mode, then perhaps triggering a release build out of hours to minimise rebuild times during development cycles. Any thoughts?

10:03 <FromGitter> <deepj> @bew Yes, I’d like to make something similar to Roda in Crystal. It’s also a way how to learn Crystal “dynamic” parts for me

10:05 <FromGitter> <deepj> Does `crystal format tool` suppor something like reading from stdin? I’m creating `crystal-fomat` package for Atom and I stuck with this because I'm heavily inspired by `elm-format`.

10:40 <FromGitter> <deepj> Hm, it seems `crystal tool format` doesn’t return proper exit statues and output on STDOUT/STDERR :worried:

11:03 <unshadow> @crisward I wouldn't do it, when in prod use --release, when in dev run without

11:19 <FromGitter> <crisward> <unshadow> Any specific reason?

11:21 <FromGitter> <luislavena> @deepj `tool format` modifies a file or directory supplied, doesn't work with stdin/stdout

11:30 <unshadow> @crisward I think that at lest for the reason of "Principle of least astonishment" , non release code is not run in production by anone, which means you might step into issues and unexpected behavior which would not happen in release (mem leaks, timeouts, etc..)

11:47 <FromGitter> <crisward> <ushadow> All my tests and crystal tests get run without --release. Production mode tends to get set by ENV vars, which I'd set to production. So in the case of crystal I'm not sure this applies.

11:55 <FromGitter> <bew> About the `crystal tool format`I agree, it would be great to have options for stdin/out!

12:03 <FromGitter> <deepj> @luislavena I’ve tried to modify formatting a given file directly but without proper exit statutes and output STDOUT/STDERR from the command it seems to be an impossible task :(

12:37 <FromGitter> <mverzilli> @crisward I don't see any downside to your strategy, especially if performance is not an issue for you. with the current compile times I'd only have a CI system compile with `--release` for me

14:32 <FromGitter> <crisward> @mverzilli Thanks. As you know, the heroku buildpack I'm uses is building very slow with --release, possibly due to 3.5 llvm (>30 minutes sometimes). Hopefully this will be resolved with the updates to the crystal omnibus build which seems to be on it's way. I don't know if this a general issue, or specific to my setup. If it was just a few minutes I'd be happy to go --release all the time.

14:42 <FromGitter> <greenbigfrog> are there any base58 shards out there? else I'll just make my own

14:56 <FromGitter> <mverzilli> not that I heard of, but have you checked http://crystalshards.xyz/?

14:57 <FromGitter> <greenbigfrog> Wow... I actually had used github search but I didn't find anything. I'll need to check if https://github.com/russ/base58 suits my needs ⏎ ty

15:17 <crystal-gh> [crystal] rockwyc992 opened pull request #4390: Compiler: fixed is_a? transform may cause an exception (master...fix/compiler/IsA_Const_enclosing_call) https://git.io/v9oVW

15:36 <FromGitter> <pglombardo> Hi all - I could use a hint/pointer on two items if possible: ⏎ ⏎ 1) How can I collect info on platform? (e.g. OS X, Linux, 64 bit etc) ⏎ 2) How can I get the command line of the running process (e.g. ARGV) [https://gitter.im/crystal-lang/crystal?at=59109080c89bb14b5ae159ae]

15:44 <FromGitter> <pglombardo> ah just found ARGV/PROGRAM_NAME in docs. 1 down :-)

15:52 <oprypin> pglombardo, https://crystal-lang.org/docs/syntax_and_semantics/compile_time_flags.html

15:53 <FromGitter> <pglombardo> Thanks - that works for me

16:14 <FromGitter> <drosehn> hmm. I wonder if it'd be interesting to implement atob/btoa in crystal (which is basically base64). Something else to play with in my copious spare time.

16:18 <FromGitter> <drosehn> getting back to @crisward 's question: it's certainly reasonable to try and see how much of a performance hit you notice (if any) by running a version compiled without `--release`. One thing I'd suggest is to make sure you're keeping track of when you're running with `--release` vs when you're running without it. It's vaguely possible that you might notice odd behavior with one version that won't happen with the

16:18 <FromGitter> ... Those differences would be "bugs", of course, but bugs do happen from time-to-time.

16:22 <zipR4ND> hey all, i'm making some progress on the graphql implementation.. see https://goo.gl/RcJgEz for the api. I have to mark StringTypes with [] because of the bug that doesn't let me check if a generic type responds to a method (resolve in this case ..) otherwise I would use ListType(MyObjectClass) .. also I'm using Module extension to turn ordinary Crystal Classes into GraphQL Object Types whie using module inclusion to define queries ..

16:23 <zipR4ND> yeah early criticism prevents me from building on the wrong foundation here ..

16:27 <FromGitter> <sdogruyol> Wow

16:27 <FromGitter> <sdogruyol> Respect

16:30 <zipR4ND> @sdogruyol thxx

16:31 <zipR4ND> the macros really help, although I'm afraid they may harm usability in production, because there are still some instabilities & unexpected behaviour ..

16:33 <zipR4ND> eg. it seems to matter, where (in which file) I extend a module in a class. for the Character class in the StarWars Schema I would have done it in star_wars_schema.cr but it only worked when I extended in star_wars_data.cr (both define the same classes, the data is just ment to define the domain while the schema defines the fields necessary to interpret it as an object type ...)

16:34 <zipR4ND> Error earlier: I have to mark ListTypes with [] not StringTypes ..

21:00 <crystal-gh> [crystal] mverzilli closed pull request #4390: Compiler: fixed is_a? transform may cause an exception (master...fix/compiler/IsA_Const_enclosing_call) https://git.io/v9oVW

22:23 <FromGitter> <elorest> Has anyone worked on getting macro’s to create formatted code so we don’t have to indent weird within macro blocks/if/else/etc to that the output looks normal?

22:38 <jandor> привет

23:18 <crystal-gh> [crystal] RX14 opened pull request #4392: Remove IO.select (master...remove-io-select) https://git.io/v9K5R