5 ms·

Building a Command-Line Application with Crystal

99 points by willlll 10y ago

throwaway7645 10y ago
Anyone compare Nim to Crystal? Both are AOT languages with syntax inspired by dynamic languages (Python & Ruby). Nim transpiles to C and Crystal uses LLVM. I've been playing with Nim, but I have a feeling Crystal will catch on quicker.
- killin_dan 10y ago
  CoC, bad compiler, bad type system, no formal proofing whatsoever, some of documentation reads like someone who knows English as a second or third or fourth language wrote it, in some cases it's just plain wrong. Elixir has already taken most of the ruby mindshare, the crystal team really isn't hacking it when it comes to presenting why crystal should be used instead of ruby except for speed. That its faster is the only viable motivation currently. I don't know anything about nim, so I can't really compare it. I wrote some compiler plugins and utilities for crystal at my company, but I ended just rewriting the component in ruby, and it runs hourly with JRuby. I hope Ruby3 gets optional types a la Dart, and a code conversion tool. Those two things would revive the entire ruby universe. Edit: last time I touched crystal was 6-7 months ago, but I doubt anything has changed. Edit edit: not an elixir fanboy. It sucks too, but at least it compiles down to BEAM so if I'm writing erlang I don't have to actually touch it as long as it works. The elixir astroturfing is out of control lately. It's cool, it ain't THAT cool.
  - _raul 10y ago
    I've started to transition from Ruby to Crystal for my pet projects and haven't hit any serious issues so far. Since you seem to have much more experience than I do, would you mind elaborating a bit on the issues you found and why you think the compiler and type system are bad?
    - Svenskunganka 10y ago
      I wouldn't count on it, judging by his previous comments he seems to prefer drive-by snipes over discussions. The fact that he made so many claims about how bad Crystal is without even attempting to explain why doesn't really make him seem credible. And the icing on the cake gotta be how he nitpicks on very minor grammatical errors made by non-native english speakers which he also seems to be completely unaware of that they are. Even I - a Swedish person - fully understand every sentence I've read so far, and English is not my primary language either. I wouldn't take this guy seriously until he brings forth some constructive criticism.
  - amorphid 10y ago
    Elixir is by far my favorite language out of the 6 I've seriously tried (Elixir, Ruby, JavaScript, Python, Java, Bash). I'm gonna dive into Go in near future.
    - name_for_now 10y ago
      I would honestly be very surprised if you said Bash was your favorite of the 6.
  - yawaramin 10y ago
    > ... some of documentation reads like someone who knows English as a second or third or fourth language wrote it, ... Um, what's wrong with that? Not everyone gets to learn English as their first language, do their efforts deserve to be discarded just because of that? I mean, I could point out that your grammar ('some of documentation') invalidates your entire comment, but that's kinda silly.
    - 35bge57dtjku 10y ago
      > I could point out that your grammar ('some of documentation') invalidates your entire comment, but that's kinda silly. He's not writing the hopefully high profile documentation for a project that people want to be used, ffs. No, it doesn't deserve to be discarded because of that, but it's clearly suboptimal.
      - yawaramin 10y ago
        You make it sound as if this version of the documentation is set in stone, forever. Can't small projects improve their documentation as they grow in popularity? E.g. do you think that Linus Torvalds' Linux documentation was always up to the highest standards of Strunk & White's prescribed English?
    - williamgb 10y ago
      I absolutely do not share the opinion of person to whom you are replying, but to state that their comment is invalid due to a trivial typo is unnecessary, and has potential to be incendiary.
      - yawaramin 10y ago
        I mean, that's why I said: > ... but that's kinda silly.
        williamgb 10y ago
        What point were you trying to make? You made one great statement, with which I agree fully, but followed with statement drawing a comparison between the documentation for a funded open-source project, and a typo in a comment thread. It's not just "kinda silly", it's utterly asinine.
  - RX14 10y ago
    >CoC I'm sorry, what does that mean? >bad compiler, bad type system I'd love to hear some feedback from you about the compiler and type system, so that we can discuss and improve them. >no formal proofing Ary has expressed interest in formally defining crystal's type system, but its understandably far down the list of priorities.
- zem 10y ago
  i think nim has better compiler options (in particular, i don't know if it has already achieved go-like cross-compilation of static binaries, but if not it's further along that road than crystal is). https://nim-lang.org/docs/nimc.html https://nim-lang.org/docs/nimc.html meanwhile, crystal does not even support windows yet. https://github.com/crystal-lang/crystal/issues/26 https://github.com/crystal-lang/crystal/issues/26
  - pjmlp 10y ago
    Go-like cross-compilation only works with pure Go libraries that don't make use of either syscalls or cgo.
    - dom96 10y ago
      I keep hearing praise over how easy it is to statically compile Go applications. Does the same apply to static compilation? Is it only easy when using pure Go libraries?
      - zem 10y ago
        according to https://dave.cheney.net/2016/01/18/cgo-is-not-go https://dave.cheney.net/2016/01/18/cgo-is-not-go > It is possible to compile a Go program entirely statically, but it is by no means simple and shows that the ramifications of including cgo in your project will ripple through your entire build and deploy life cycle. (that article also contains the beautiful line "You’re not writing a Go program that uses some logic from a C library, instead you’re writing a Go program that has to coexist with a belligerent piece of C code that is hard to replace, has the upper hand negotiations, and doesn’t care about your problems.")
      - pjmlp 10y ago
        Usually those praises come from young developers that never developed software back in the days when static linking was the only option. So they look at Go's static linking, their experiences using -static in gcc and glibc, and think they just found something magical. It is interesting the amount of knowledge that gets lost within a few generations.
name_for_now 10y ago
Hmm, seems like the author either does not know about JSON.mapping [0] or just chose not to use it. Honestly, the corresponding macro for YAML is the main reason I use Crystal for CLI apps since it comes with the stdlib. YAML is easy enough to write quickly that I don't have to do too much with option parsing. Plus you can just use your editor rather than the shell. [0] https://crystal-lang.org/api/master/JSON.html#mapping%28properties%2Cstrict%3Dfalse%29-macro https://crystal-lang.org/api/master/JSON.html#mapping%28prop...
- jclem 10y ago
  I do use it in this project, but most definitely in the wrong way. I had a little trouble seeing how I could use it for getting at nested values of varying types and still have some sane error handling. This post is very much about my first pass at this and my first experience with Crystal. Cleaning up the JSON parsing is the next item on my list when I have time.
  - name_for_now 10y ago
    You can pass union types to the macro, so that should handle your "varying types" problem. In terms of nesting, I just create another struct with its own JSON.mapping and use that as the type for the field.
  - name_for_now 10y ago
    Here's a quick gist that does what I think you want, modulo any specific error handling. https://gist.github.com/usernamefornow/14f972df9429146b46846595b80e5fe2 https://gist.github.com/usernamefornow/14f972df9429146b46846... edit: note that the cleanest way would probably to write your own converter so you could avoid the union type, but I didn't want to put that in and have to write a comment about it.
    - jclem 10y ago
      Makes sense, thanks. For some reason, I was set on parsing the "env" values into instances of the same class, regardless of what type they were in JSON, but it's not really necessary. I'll also look into writing custom converters, because I'll likely need that in the future.
neovintage 10y ago
I've spent a number of years using ruby and I got super excited about the potential of Crystal about a year ago. The notion that I can use a Ruby-like syntax and ship a binary to deploy my web apps was killer. But, a year ago the language was a bit rough, and to some extent, it still is. The team behind it is open about their goals and it blows me away with how much changes on a daily basis for a team of their size. Hopefully, the crystal team can set out to complete what they set their sights on for 2017[1]. That would really round out the feature set for a 1.0 release. I think what I like the most is the community is willing to help you out. Although, nowadays a strong community is what makes or breaks a language. [1] https://crystal-lang.org/2016/12/29/crystal-new-year-resolutions-for-2017-1-0.html https://crystal-lang.org/2016/12/29/crystal-new-year-resolut...
grogenaut 10y ago
Dangit, doesn't work natively on windows without bash :( And I'm all windows lately. Guess I can try it on my mac but that's like over there. I do love the multi-platform-ness and single file of go, so it'll be cool to try this and get some ruby love.
- dejv 10y ago
  It seems like porting Crystal to run on WIndows is work in progress: https://github.com/crystal-lang/crystal/pull/3582 https://github.com/crystal-lang/crystal/pull/3582
sdogruyol 10y ago
BTW the Crystal website has just been redesigned :) https://crystal-lang.org/ https://crystal-lang.org/
devmunchies 10y ago
I really, really like crystal but was bitten a few times on a project because it is still not quite ready for primetime. Because its so young, there have been a handful of breaking changes which affected my code a few times (and even worse, my dependencies). They have a good concurrency model but I'm waiting for them to add parallelism. I mostly just crave the solidarity in tooling, dependencies, APIs, compiler options/performance which will come after hitting 1.0 (i hope). I'm excited for the day I can go all in. Keep it up Crystal team!

99 points by willlll 10y ago

Anyone compare Nim to Crystal? Both are AOT languages with syntax inspired by dynamic languages (Python & Ruby). Nim transpiles to C and Crystal uses LLVM. I've been playing with Nim, but I have a feeling Crystal will catch on quicker.

killin_dan 10y ago
CoC, bad compiler, bad type system, no formal proofing whatsoever, some of documentation reads like someone who knows English as a second or third or fourth language wrote it, in some cases it's just plain wrong. Elixir has already taken most of the ruby mindshare, the crystal team really isn't hacking it when it comes to presenting why crystal should be used instead of ruby except for speed. That its faster is the only viable motivation currently. I don't know anything about nim, so I can't really compare it. I wrote some compiler plugins and utilities for crystal at my company, but I ended just rewriting the component in ruby, and it runs hourly with JRuby. I hope Ruby3 gets optional types a la Dart, and a code conversion tool. Those two things would revive the entire ruby universe. Edit: last time I touched crystal was 6-7 months ago, but I doubt anything has changed. Edit edit: not an elixir fanboy. It sucks too, but at least it compiles down to BEAM so if I'm writing erlang I don't have to actually touch it as long as it works. The elixir astroturfing is out of control lately. It's cool, it ain't THAT cool.
- _raul 10y ago
  I've started to transition from Ruby to Crystal for my pet projects and haven't hit any serious issues so far. Since you seem to have much more experience than I do, would you mind elaborating a bit on the issues you found and why you think the compiler and type system are bad?
  - Svenskunganka 10y ago
    I wouldn't count on it, judging by his previous comments he seems to prefer drive-by snipes over discussions. The fact that he made so many claims about how bad Crystal is without even attempting to explain why doesn't really make him seem credible. And the icing on the cake gotta be how he nitpicks on very minor grammatical errors made by non-native english speakers which he also seems to be completely unaware of that they are. Even I - a Swedish person - fully understand every sentence I've read so far, and English is not my primary language either. I wouldn't take this guy seriously until he brings forth some constructive criticism.
- amorphid 10y ago
  Elixir is by far my favorite language out of the 6 I've seriously tried (Elixir, Ruby, JavaScript, Python, Java, Bash). I'm gonna dive into Go in near future.
  - name_for_now 10y ago
    I would honestly be very surprised if you said Bash was your favorite of the 6.
- yawaramin 10y ago
  > ... some of documentation reads like someone who knows English as a second or third or fourth language wrote it, ... Um, what's wrong with that? Not everyone gets to learn English as their first language, do their efforts deserve to be discarded just because of that? I mean, I could point out that your grammar ('some of documentation') invalidates your entire comment, but that's kinda silly.
  - 35bge57dtjku 10y ago
    > I could point out that your grammar ('some of documentation') invalidates your entire comment, but that's kinda silly. He's not writing the hopefully high profile documentation for a project that people want to be used, ffs. No, it doesn't deserve to be discarded because of that, but it's clearly suboptimal.
    - yawaramin 10y ago
      You make it sound as if this version of the documentation is set in stone, forever. Can't small projects improve their documentation as they grow in popularity? E.g. do you think that Linus Torvalds' Linux documentation was always up to the highest standards of Strunk & White's prescribed English?
  - williamgb 10y ago
    I absolutely do not share the opinion of person to whom you are replying, but to state that their comment is invalid due to a trivial typo is unnecessary, and has potential to be incendiary.
    - yawaramin 10y ago
      I mean, that's why I said: > ... but that's kinda silly.
      - williamgb 10y ago
        What point were you trying to make? You made one great statement, with which I agree fully, but followed with statement drawing a comparison between the documentation for a funded open-source project, and a typo in a comment thread. It's not just "kinda silly", it's utterly asinine.
- RX14 10y ago
  >CoC I'm sorry, what does that mean? >bad compiler, bad type system I'd love to hear some feedback from you about the compiler and type system, so that we can discuss and improve them. >no formal proofing Ary has expressed interest in formally defining crystal's type system, but its understandably far down the list of priorities.
zem 10y ago
i think nim has better compiler options (in particular, i don't know if it has already achieved go-like cross-compilation of static binaries, but if not it's further along that road than crystal is). https://nim-lang.org/docs/nimc.html https://nim-lang.org/docs/nimc.html meanwhile, crystal does not even support windows yet. https://github.com/crystal-lang/crystal/issues/26 https://github.com/crystal-lang/crystal/issues/26
- pjmlp 10y ago
  Go-like cross-compilation only works with pure Go libraries that don't make use of either syscalls or cgo.
  - dom96 10y ago
    I keep hearing praise over how easy it is to statically compile Go applications. Does the same apply to static compilation? Is it only easy when using pure Go libraries?
    - zem 10y ago
      according to https://dave.cheney.net/2016/01/18/cgo-is-not-go https://dave.cheney.net/2016/01/18/cgo-is-not-go > It is possible to compile a Go program entirely statically, but it is by no means simple and shows that the ramifications of including cgo in your project will ripple through your entire build and deploy life cycle. (that article also contains the beautiful line "You’re not writing a Go program that uses some logic from a C library, instead you’re writing a Go program that has to coexist with a belligerent piece of C code that is hard to replace, has the upper hand negotiations, and doesn’t care about your problems.")
    - pjmlp 10y ago
      Usually those praises come from young developers that never developed software back in the days when static linking was the only option. So they look at Go's static linking, their experiences using -static in gcc and glibc, and think they just found something magical. It is interesting the amount of knowledge that gets lost within a few generations.

name_for_now 10y ago

Hmm, seems like the author either does not know about JSON.mapping [0] or just chose not to use it. Honestly, the corresponding macro for YAML is the main reason I use Crystal for CLI apps since it comes with the stdlib. YAML is easy enough to write quickly that I don't have to do too much with option parsing. Plus you can just use your editor rather than the shell. [0] https://crystal-lang.org/api/master/JSON.html#mapping%28properties%2Cstrict%3Dfalse%29-macro https://crystal-lang.org/api/master/JSON.html#mapping%28prop...

jclem 10y ago
I do use it in this project, but most definitely in the wrong way. I had a little trouble seeing how I could use it for getting at nested values of varying types and still have some sane error handling. This post is very much about my first pass at this and my first experience with Crystal. Cleaning up the JSON parsing is the next item on my list when I have time.
- name_for_now 10y ago
  You can pass union types to the macro, so that should handle your "varying types" problem. In terms of nesting, I just create another struct with its own JSON.mapping and use that as the type for the field.
- name_for_now 10y ago
  Here's a quick gist that does what I think you want, modulo any specific error handling. https://gist.github.com/usernamefornow/14f972df9429146b46846595b80e5fe2 https://gist.github.com/usernamefornow/14f972df9429146b46846... edit: note that the cleanest way would probably to write your own converter so you could avoid the union type, but I didn't want to put that in and have to write a comment about it.
  - jclem 10y ago
    Makes sense, thanks. For some reason, I was set on parsing the "env" values into instances of the same class, regardless of what type they were in JSON, but it's not really necessary. I'll also look into writing custom converters, because I'll likely need that in the future.

neovintage 10y ago

I've spent a number of years using ruby and I got super excited about the potential of Crystal about a year ago. The notion that I can use a Ruby-like syntax and ship a binary to deploy my web apps was killer. But, a year ago the language was a bit rough, and to some extent, it still is. The team behind it is open about their goals and it blows me away with how much changes on a daily basis for a team of their size. Hopefully, the crystal team can set out to complete what they set their sights on for 2017[1]. That would really round out the feature set for a 1.0 release. I think what I like the most is the community is willing to help you out. Although, nowadays a strong community is what makes or breaks a language. [1] https://crystal-lang.org/2016/12/29/crystal-new-year-resolutions-for-2017-1-0.html https://crystal-lang.org/2016/12/29/crystal-new-year-resolut...

grogenaut 10y ago

Dangit, doesn't work natively on windows without bash :( And I'm all windows lately. Guess I can try it on my mac but that's like over there. I do love the multi-platform-ness and single file of go, so it'll be cool to try this and get some ruby love.

dejv 10y ago
It seems like porting Crystal to run on WIndows is work in progress: https://github.com/crystal-lang/crystal/pull/3582 https://github.com/crystal-lang/crystal/pull/3582

sdogruyol 10y ago

BTW the Crystal website has just been redesigned :) https://crystal-lang.org/ https://crystal-lang.org/

devmunchies 10y ago

I really, really like crystal but was bitten a few times on a project because it is still not quite ready for primetime. Because its so young, there have been a handful of breaking changes which affected my code a few times (and even worse, my dependencies). They have a good concurrency model but I'm waiting for them to add parallelism. I mostly just crave the solidarity in tooling, dependencies, APIs, compiler options/performance which will come after hitting 1.0 (i hope). I'm excited for the day I can go all in. Keep it up Crystal team!