Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Repo info
Activity
Elia Schito
@elia
@catmando ok, that could be a collateral effect of the improved behavior, @sfcgeorge let me know what's that, if it's an actual issue I'd like to look into it
Mitch VanDuyn
@catmando
@elia is there any way to run try opal on older versions?
Simon George
@sfcgeorge

A version selector on Try Opal would be amazingly helpful.

In this case I just know that making the second parameter true made something work in 0.11 that had worked in 0.10. I don't know which one the bug is in though, just because it worked before doesn't mean it was working as intended.

Elia Schito
@elia
@catmando no, but you can run the CLI on any installed version with opal _0.10.5_ -e 'p RUBY_ENGINE_VERSION'
@sfcgeorge yeah, 0.11 catched a bunch of errors in one of my apps
Mitch VanDuyn
@catmando
@elia :-) I have never used the CLI
I will try it now :-)
is there an opal irb ?
Mitch VanDuyn
@catmando
@elia - just using some simple tests I can see no behavior change...
Elia Schito
@elia
@catmando there's opal-repl, it needs thrubyracer IIRC
G. Gibson
@mistergibson
@elia I'm having trouble adding methods to class String - is this unsupported?
Elia Schito
@elia
should work fine, unless you're trying to mutate the string, inheriting from String it's more of a hackery but still supported
G. Gibson
@mistergibson
no, it mainly is for some methods that do regex matching
Elia Schito
@elia
@mistergibson can you paste your attempt here?
G. Gibson
@mistergibson
yeah, I'll whip up a gist for you
Elia Schito
@elia
Looks good, what kind of error are you getting?
G. Gibson
@mistergibson
This: uncaught exception: TypeError: can't convert String into Kernel::String (String#to_str gives String
I was surprised
Elia Schito
@elia
Seems to be working here, p "AsdfAasdf".split_camelcase resulted in ["Asdf", "Aasdf"]
G. Gibson
@mistergibson
I wonder what the deal is?
it won't even accept my method defs
it must be in the way I'm building the bootstrap
Elia Schito
@elia
maybe it's some weird interaction due to all those requires, try removing most of them and see if it works, if it does then you can start looking for the one to blame
G. Gibson
@mistergibson
sure
G. Gibson
@mistergibson
Hrm ... loaded it right after 'opal/full' and got the same error - never makes it to the other requires.
Elia Schito
@elia
have you tried just with 'opal'?
G. Gibson
@mistergibson
That's my next stop :)
Same error
so .... require 'opal' ... then the .rb file on require_remote gives the error.
Elia Schito
@elia
gotta go to sleep :crescent_moon: , I'll read tomorrow :wave:
G. Gibson
@mistergibson
ok ... sleep tight :)
G. Gibson
@mistergibson
@elia -- when you get up: When I exclude 'opal-parser' I don't get the error.
Ghost
@ghost~59874280d73408ce4f704e9c

@elia i think there is a bug in js inlining, sometimes not generating a necessary return
example opal-active support String.camelize
non-working, generated by 0.11:

return (Opal.defn(self, '$camelize', TMP_String_camelize_2 = function $$camelize(first_letter) {
      var self = this;

      if (first_letter == null) {
        first_letter = "upper";
      }
      self.$underscore().replace(/(^|_)([^_]+)/g, function(match, pre, word, index) {
      var capitalize = first_letter === "upper" || index > 0;
      return capitalize ? word.substr(0,1).toUpperCase()+word.substr(1) : word;
    })

manual workaround:

    return (Opal.defn(self, '$camelize', TMP_String_camelize_2 = function $$camelize(first_letter) {
      var self = this;

      if (first_letter == null) {
        first_letter = "upper";
      }
      return self.$underscore().replace(/(^|_)([^_]+)/g, function(match, pre, word, index) {
      var capitalize = first_letter === "upper" || index > 0;
      return capitalize ? word.substr(0,1).toUpperCase()+word.substr(1) : word;
    })
note the added return in the middle
original code:
  def camelize(first_letter = :upper)
    `#{underscore}.replace(/(^|_)([^_]+)/g, function(match, pre, word, index) {
      var capitalize = #{first_letter} === #{:upper} || index > 0;
      return capitalize ? word.substr(0,1).toUpperCase()+word.substr(1) : word;
    })`
  end
workaround:
  def camelize(first_letter = :upper)
    `return #{underscore}.replace(/(^|_)([^_]+)/g, function(match, pre, word, index) {
      var capitalize = #{first_letter} === #{:upper} || index > 0;
      return capitalize ? word.substr(0,1).toUpperCase()+word.substr(1) : word;
    })`
  end
Ghost
@ghost~59874280d73408ce4f704e9c
now, where would i have to look for fixing this in the compiler?
Ghost
@ghost~59874280d73408ce4f704e9c
or how can i debug such things?
Simon George
@sfcgeorge
That native inlining issue @janbiedermann mentions is reported here opal/opal-activesupport#16 - Another workaround is rather than adding a return you can just call .to_s on the backtick string. I don't know why that would change what is happening. Does it force a native value to be cast into an Opal value or something?
Elia Schito
@elia

@janbiedermann try adding a semicolon before closing the backticks, that said to debug this kind of stuff I often do

opal -cEO -e "def foo; `return 123, function(){}` end"

and

opal --sexp  -e "def foo; `return 123, function(){}` end"
Ghost
@ghost~59874280d73408ce4f704e9c
@sfcgeorge when you add the the .to_s the compiler adds the returnabove, thats why it works with .to_s
Simon George
@sfcgeorge
@janbiedermann Interesting. That almost makes it seem like expected behaviour. If you show you intend to use the value then it returns it. Seems unhelpful though, especially if you don't know what the return type will be.
Ghost
@ghost~59874280d73408ce4f704e9c
@elia adding a ;doesnt help
Ghost
@ghost~59874280d73408ce4f704e9c

@iliabylich changing https://github.com/opal/opal/blob/ee50768d368959b33a77b58028a90ec9bded64a1/lib/opal/compiler.rb#L436 to

          if multiline
            # xstr starts with interpolation
            # then it must contain js_return inside
            if first_child.type == :begin
              s(:js_return, sexp)
            else
              sexp
            end
          else
            if first_child.type == :str

helps the above troubled case, also tests pass, though i am not sure, if that is the right thing to do there, looks too specific, or would you be ok with that? then i would provide a PR

G. Gibson
@mistergibson
Question: is there any plan to support REXML in the StandardLib? I notice it is missing - compared to MRI.
Ghost
@ghost~59874280d73408ce4f704e9c

@elia @iliabylich found another issue, for:

      def native_state_changed?(next_state_hash)
        next_state = next_state_hash.to_n
        %x{
          var current_state = #{@native}.state

compiler generates:

    next_state = next_state_hash.$to_n();
    return
    var current_state = self["native"].state

which wont work. Now fixed it here https://github.com/opal/opal/blob/ee50768d368959b33a77b58028a90ec9bded64a1/lib/opal/compiler.rb#L446 with this:

            if first_child.type == :str
              old_value = first_child.children[0]
              if old_value.include?('return')
                # 'return' is already there
                sexp
              else
                second_child, *second_rest_children = *rest_children
                if second_child && second_child.type == :str && !(old_value =~ /^\s*\n$/).nil? && !(second_child.children[0] =~ /^\s*var/).nil?
                  sexp
                else
                  first_child = s(:js_return, first_child)
                  sexp.updated(nil, [first_child, *rest_children])
                end
              end
works, tests work, but it looks so ugly. How can i make that better?
Ilya Bylich
@iliabylich
Opal injects return only for a single line x-strings. But the criteria for marking x-string as multiline is wrong - multiline = strs.any? { |str| str.end_with?(";\n") }
So if there are no lines ending with ; Opal decides that that's a single line x-string and injects a return if x-string doesn't have it