Unknown variables 2
Locked
 
|
Hello,
the `(smalltalk.Foo || Foo)` (and unknown variable as invitation to use global variables seamlessly; which is a bit of an antipattern IMHO) is still making me nervous. So I came with the different proposal: Let <...> have two uses: 1. Inline JS statement (as its node name suggests). Every <...> will be parsed as this, except: 2. Global variable access. If inside <...> is an identifier and nothing more, it should be parsed as ExternalGlobalVariableNode (a new node). So `console log: 'Hello, world'` gets only just a little more complicated (and more explicit): `<console> log: 'Hello, world!'`. More over, any `Capitalized` alone would be unambiguously interpret as smalltalk.Capitalized and to use outer variable beginning with capital letter, one should use eg. `<Math> max: 2 and: 3 and: 4`. Is this more acceptable? Herby |
|
|
Herby Vojčík <[hidden email]> writes:
Hello Herby, This indeed sounds better to me :) I still have a few remarks: - I don't like the idea of having the same syntax for 2 different semantics, even more because it's not 'normal' Smalltalk syntax - instead of having a special syntax, what about adding some kind of simple namespace for javascript object access? Something like the following: JavaScript console log: 'hello' It looks a lot more Smalltalkish, and instead of adding new semantic, we can add a new pseudo variable. Cheers, Nico > Hello, > > the `(smalltalk.Foo || Foo)` (and unknown variable as invitation to > use global variables seamlessly; which is a bit of an antipattern > IMHO) is still making me nervous. So I came with the different > proposal: > > Let <...> have two uses: > 1. Inline JS statement (as its node name suggests). Every <...> will > be parsed as this, except: > 2. Global variable access. If inside <...> is an identifier and > nothing more, it should be parsed as ExternalGlobalVariableNode (a new > node). > > So `console log: 'Hello, world'` gets only just a little more > complicated (and more explicit): `<console> log: 'Hello, > world!'`. More over, any `Capitalized` alone would be unambiguously > interpret as smalltalk.Capitalized and to use outer variable beginning > with capital letter, one should use eg. `<Math> max: 2 and: 3 and: 4`. > > Is this more acceptable? > > Herby |
|
|
JavaScript console log: 'hello'
+1 On Sun, May 6, 2012 at 2:31 PM, <[hidden email]> wrote: Herby Vojčík <[hidden email]> writes: |
|
|
In reply to this post by Nicolas Petton
[hidden email] wrote:
> Herby Vojčík<[hidden email]> writes: > > Hello Herby, > > This indeed sounds better to me :) > I still have a few remarks: > > - I don't like the idea of having the same syntax for 2 different > semantics, even more because it's not 'normal' Smalltalk syntax Well, if <> would only be standalone JS statement, I'd like it. > - instead of having a special syntax, what about adding some kind of > simple namespace for javascript object access? Something like the > following: > > JavaScript console log: 'hello' Javascript isn't written in two words :-( You do to JS what others do to Smalltalk by writing SmallTalk (this one is scarier, of course). > It looks a lot more Smalltalkish, and instead of adding new semantic, we > can add a new pseudo variable. I would call it Host. What if Amber would run on Dart / Clojure / godknowswhatbrowserlanguage in the future? > Cheers, > Nico > >> Hello, >> >> the `(smalltalk.Foo || Foo)` (and unknown variable as invitation to >> use global variables seamlessly; which is a bit of an antipattern >> IMHO) is still making me nervous. So I came with the different >> proposal: >> >> Let<...> have two uses: >> 1. Inline JS statement (as its node name suggests). Every<...> will >> be parsed as this, except: >> 2. Global variable access. If inside<...> is an identifier and >> nothing more, it should be parsed as ExternalGlobalVariableNode (a new >> node). >> >> So `console log: 'Hello, world'` gets only just a little more >> complicated (and more explicit): `<console> log: 'Hello, >> world!'`. More over, any `Capitalized` alone would be unambiguously >> interpret as smalltalk.Capitalized and to use outer variable beginning >> with capital letter, one should use eg. `<Math> max: 2 and: 3 and: 4`. >> >> Is this more acceptable? >> >> Herby |
|
|
Herby Vojčík <[hidden email]> writes:
> [hidden email] wrote: >> Herby Vojčík<[hidden email]> writes: >> >> Hello Herby, >> >> This indeed sounds better to me :) >> I still have a few remarks: >> >> - I don't like the idea of having the same syntax for 2 different >> semantics, even more because it's not 'normal' Smalltalk syntax > > Well, if <> would only be standalone JS statement, I'd like it. > >> - instead of having a special syntax, what about adding some kind of >> simple namespace for javascript object access? Something like the >> following: >> >> JavaScript console log: 'hello' > > Javascript isn't written in two words :-( You do to JS what others do > to Smalltalk by writing SmallTalk (this one is scarier, of course). Oh, really? I always thought it was JavaScript, not Javascript. Wikipedia and the MDN seem to agree with me :) https://developer.mozilla.org/en/JavaScript > >> It looks a lot more Smalltalkish, and instead of adding new semantic, we >> can add a new pseudo variable. > > I would call it Host. What if Amber would run on Dart / Clojure / > godknowswhatbrowserlanguage in the future? Then we'll change it :) Amber is tightly coupled to Java{S,s}cript anyway. Cheers, Nico >> Cheers, >> Nico >> >>> Hello, >>> >>> the `(smalltalk.Foo || Foo)` (and unknown variable as invitation to >>> use global variables seamlessly; which is a bit of an antipattern >>> IMHO) is still making me nervous. So I came with the different >>> proposal: >>> >>> Let<...> have two uses: >>> 1. Inline JS statement (as its node name suggests). Every<...> will >>> be parsed as this, except: >>> 2. Global variable access. If inside<...> is an identifier and >>> nothing more, it should be parsed as ExternalGlobalVariableNode (a new >>> node). >>> >>> So `console log: 'Hello, world'` gets only just a little more >>> complicated (and more explicit): `<console> log: 'Hello, >>> world!'`. More over, any `Capitalized` alone would be unambiguously >>> interpret as smalltalk.Capitalized and to use outer variable beginning >>> with capital letter, one should use eg. `<Math> max: 2 and: 3 and: 4`. >>> >>> Is this more acceptable? >>> >>> Herby |
|
|
[hidden email] wrote: > Herby Vojčík<[hidden email]> writes: > >> [hidden email] wrote: >>> Herby Vojčík<[hidden email]> writes: >>> >>> Hello Herby, >>> >>> This indeed sounds better to me :) >>> I still have a few remarks: >>> >>> - I don't like the idea of having the same syntax for 2 different >>> semantics, even more because it's not 'normal' Smalltalk syntax >> Well, if<> would only be standalone JS statement, I'd like it. >> >>> - instead of having a special syntax, what about adding some kind of >>> simple namespace for javascript object access? Something like the >>> following: >>> >>> JavaScript console log: 'hello' >> Javascript isn't written in two words :-( You do to JS what others do >> to Smalltalk by writing SmallTalk (this one is scarier, of course). > > Oh, really? I always thought it was JavaScript, not > Javascript. Wikipedia and the MDN seem to agree with me :) > https://developer.mozilla.org/en/JavaScript |
|
|
In reply to this post by Nicolas Petton
[hidden email] wrote:
> Herby Vojčík<[hidden email]> writes: > > Hello Herby, > > This indeed sounds better to me :) > I still have a few remarks: > > - I don't like the idea of having the same syntax for 2 different > semantics, even more because it's not 'normal' Smalltalk syntax > > - instead of having a special syntax, what about adding some kind of > simple namespace for javascript object access? Something like the > following: > > JavaScript console log: 'hello' This has problem that it does not solve capitalized globals (unless I am mistaken, message can't start with capital letter). The possible workarounds I see are (for Math example): 'Math' jsGlobal JavaScript @ 'Math' The latter must be parenthesized in some situation anyway, JavaScript at: 'Math' is quirky because it must be parenthesized anyway. And one more important thing: you cannot assign to 'JavaScript console'. Only thing you can do is "JavaScript at: 'console' put: ...". So either at:put: (hardly inlineable) or some sort of real namespace (JavaScript.console, Javascript.Math without whitespace after dot; but Pharo does not have namespaces, does it?) or I would revitalize my proposal <identifier> for any host-provided identifier. > It looks a lot more Smalltalkish, and instead of adding new semantic, we > can add a new pseudo variable. > > Cheers, > Nico > >> Hello, >> >> the `(smalltalk.Foo || Foo)` (and unknown variable as invitation to >> use global variables seamlessly; which is a bit of an antipattern >> IMHO) is still making me nervous. So I came with the different >> proposal: >> >> Let<...> have two uses: >> 1. Inline JS statement (as its node name suggests). Every<...> will >> be parsed as this, except: >> 2. Global variable access. If inside<...> is an identifier and >> nothing more, it should be parsed as ExternalGlobalVariableNode (a new >> node). >> >> So `console log: 'Hello, world'` gets only just a little more >> complicated (and more explicit): `<console> log: 'Hello, >> world!'`. More over, any `Capitalized` alone would be unambiguously >> interpret as smalltalk.Capitalized and to use outer variable beginning >> with capital letter, one should use eg. `<Math> max: 2 and: 3 and: 4`. >> >> Is this more acceptable? >> >> Herby |
|
|
Herby Vojčík <[hidden email]> writes:
> This has problem that it does not solve capitalized globals (unless I > am mistaken, message can't start with capital letter). Yep, messages can start with a capital letter. Nico |
|
|
[hidden email] wrote: > Herby Vojčík<[hidden email]> writes: > >> This has problem that it does not solve capitalized globals (unless I >> am mistaken, message can't start with capital letter). > > Yep, messages can start with a capital letter. Ah, ok. And writing is also ok via JavaScript Foo: bar, I have somehow forgot, sorry. Ok, then. I must ask Brendan Eich how is Javascript really called, to be sure :-) > Nico Herby |
«
Return to Amber Smalltalk
|
1 view|%1 views
| Free forum by Nabble | Edit this page |