07:32 <FromGitter> <choallin> Is there a way in Crystal to instantiate an Object from a String? Something like ```string_from_trustworthy_source.constantize.new```

08:09 <FromGitter> <overnet> Hi @drosehn, thank you for the reply. ⏎ Could you please give me an example how would you split this regexes? ⏎ ⏎ 1) some_keyword(value_1|value_2|value_3)some_keyword_2(value_4|value_5|value_6)some_keyword_3 ⏎ 2) some_keyword(value_1|value_2|.....value_10000)some_keyword_2(value_4|value_5|.....value_200000)some_keyword_3 ... [https://gitter.im/crystal-lang/crystal?at=582d65a0e097df7575c14d48]

12:03 <FromGitter> <raydf> @overnet, is this for full text search?

12:43 <FromGitter> <luislavena> @choallin no, as a compiled language, types needs to be known at compilation time. You can workaround that by using a `case` against the string and instantiate that constant, or build a hash as mapping for those.

12:54 <crystal-gh> [crystal] asterite pushed 3 new commits to master: https://git.io/vXHag

12:54 <crystal-gh> crystal/master be619da Ary Borenszweig: Fixed #3523: only use short block notation when stringifying a call if the arguments are the ones reserved by the language (like __arg0)

12:54 <crystal-gh> crystal/master df20bdc Ary Borenszweig: Added `@def` inside macros that takes the value of the current method. Fixes #1582

12:54 <crystal-gh> crystal/master f2854ea Ary Borenszweig: Add __END_LINE__ magic constant in default argument value. Fixes #3552

13:13 <FromGitter> <overnet> @raydf, no, there is a description of what I am trying to implement in the yesterday's posts.

13:13 <travis-ci> crystal-lang/crystal#be619da (master - Fixed #3523: only use short block notation when stringifying a call if the arguments are the ones reserved by the language (like __arg0)): The build was broken. https://travis-ci.org/crystal-lang/crystal/builds/176689139

13:13 <DeBot> https://github.com/crystal-lang/crystal/issues/3523 (Macro yields are producing invalid code)

13:22 <FromGitter> <asterite> @overnet What are you trying to implement? What's the program?

13:30 <crystal-gh> [crystal] asterite pushed 1 new commit to master: https://git.io/vXH6f

13:30 <crystal-gh> crystal/master 381b4fd Ary Borenszweig: Reformat file

13:35 <FromGitter> <overnet> @asterite read posts from yesterday, the system that I build is very complex, it is difficult to describe shortly, it is currently built with ruby, but ruby is slow in processing large regexes, I have tried crystal but crystal does not allow to use lage regexes.

13:36 <FromGitter> <luislavena> @asterite https://gitter.im/crystal-lang/crystal?at=582c78e531c5cbef43036a24

13:36 <FromGitter> <luislavena> appears that final regex is 2.5MB size (or something like that)

13:37 <FromGitter> <luislavena> AFAIK default setting of PCRE is way smaller than that :smile:

13:42 <BlaXpirit> overnet, why would Ruby be slow in processing large regexes? there's no reason for that. maybe you should just listen to what people are saying and realize that regex is actually not usable here

13:53 <FromGitter> <overnet> @BlaXpirit have you tried to find elements from the array of strings by very, very long regex in ruby when the array contains 100000 elements? I show some samples above, could you show me please an example of how to resize these regexes, or tell me what is usable here?

13:56 <travis-ci> crystal-lang/crystal#381b4fd (master - Reformat file): The build was fixed. https://travis-ci.org/crystal-lang/crystal/builds/176698005

14:32 <crystal-gh> [crystal] asterite pushed 1 new commit to master: https://git.io/vXHyh

14:32 <crystal-gh> crystal/master 53be65d Tatsujin Chin: Add intro sort to Array (#3514)

14:37 <FromGitter> <asterite> @overnet What I mean is, if you are using such long regexes there might be a chance that it's not the correct solution for the problem

14:38 <FromGitter> <asterite> So we need to know the problem first to know if we really need a long regex as a solution

15:26 <travis-ci> crystal-lang/crystal#53be65d (master - Add intro sort to Array (#3514)): The build passed. https://travis-ci.org/crystal-lang/crystal/builds/176714057

15:26 <DeBot> https://github.com/crystal-lang/crystal/pull/3514 (Add intro sort to Array)

17:48 <maxpowa> Is there a way to use a char as a type in a yaml mapping? I seem to be getting "undefined method 'new' for Char:Class" if I try to use it as the type in the mapping...

17:50 <maxpowa> example: https://carc.in/#/r/1e9q

17:51 <RX14> maxpowa, custom converter it seems

17:51 <maxpowa> ok :V

17:51 <RX14> you can PR it

17:52 <RX14> or monkeypatch it in

18:25 <FromGitter> <drosehn> @overnet -- I'd need to know more context before I'd say how I'd approach the problem. For instance, on the project where I talked about "matching the middle in one regex, and then processing the `lhs` and `rhs` afterwards", that worked because the middle regex was guaranteed to be distinctive. It was a very effective solution, but only because I knew all about the specific data I needed to parse.

18:28 <FromGitter> <drosehn> It sounds like you are tackling a large project (which is certainly a good thing to try to do in crystal), but we're not going to know what tactics to suggest if we're looking at a tiny part of the total problem. I've lived through this before, on both sides of a chat-discussion. All that happens is everyone (on both sides) will get frustrated.

18:31 <FromGitter> <drosehn> "distinctive" in the sense that if (for example) the middle was the regex `(fred | barney | wilma | betty)`, I was absolutely certain that none of those words would show up in either the left-hand-side or right-hand-side of the data line.

18:48 <FromGitter> <redcodefinal> Morning Crystal Bros

19:04 <FromGitter> <sdogruyol> @redcodefinal morning / evening

20:33 <FromGitter> <jots_twitter> @overnet if you can normalize your data (strip and squeeze whitespace, downcase etc) then I think this will work much faster for you then a regex: ⏎ (if i'm understanding your problem correctly.) ⏎ ⏎ ```code paste, see link``` ⏎ ... [https://gitter.im/crystal-lang/crystal?at=582e1429e51081951edb93b3]

20:52 <FromGitter> <johnjansen> @sdogruyol anything like a HAML that works with kemal

21:06 <FromGitter> <jots_twitter> @johnjansen there is slang which appears similar: https://github.com/jeromegn/slang

21:07 <FromGitter> <sdogruyol> @johnjansen slang

21:07 <FromGitter> <johnjansen> thanks @jots_twitter

21:09 <FromGitter> <drosehn> @jots_twitter ... okay, how do you get gitter to show a block of code-text like that? three "`"s followed by cr?

21:09 <FromGitter> <johnjansen> yeps ;-)

21:10 <FromGitter> <jots_twitter> yes and then end it with three backticks

21:10 <FromGitter> <johnjansen> three backticks new line code new line three backticks

21:10 <FromGitter> <drosehn> Okay. thanks. that seemed likely, but I didn't want to try it and then find out I made the wrong guess!

21:16 <FromGitter> <drosehn> @overnet But to answer the question you asked: **Iff* I was **certain** that `some_keyword`, `some_keyword_2` and `some_keyword_3` could never show up except at the spots you listed, and I'll assume there is some delimiter around the keywords (such as spaces), I would:

21:16 <FromGitter> <drosehn> ```left_vals = $~[2] ⏎ right_vals = $~[4]``` ⏎ ⏎ [hrm. that didn't work!] [https://gitter.im/crystal-lang/crystal?at=582e1e43004785313cbb2917]

21:17 <FromGitter> <drosehn> [no, wait, maybe it did]

21:17 <FromGitter> <drosehn> I'd also save copies of `$~[1]`, `$~[3]` and `$~[5]` if there was more than one keyword possible for each "keyword position", and I needed to keep track of which keywords matched. That way, the regex's needed to process `left_vals` and `right_vals` could be considerably smaller than if I have everything in one huge regex. But I suspect that you'll look at this and say *"that won't work for me, due to details which I haven't told

21:17 <FromGitter> ... anyone yet!".* That's what will get both you and us frustrated.

21:21 <FromGitter> <drosehn> ignore the initial "cr" before the "dataline"...

21:24 <FromGitter> <drosehn> ```code paste, see link``` [https://gitter.im/crystal-lang/crystal?at=582e1ffbc3afa78a1e1ff0a9]

21:24 <FromGitter> <drosehn> Okay, that worked better!