Variable names / inline JS: proposals of stricter method syntax

nicolas petton wrote:

I fully agree :) A few changes in the parser and compiler should do the
trick.

Which one? I like the first, it's cleaner.

Herby

Nico

2012/5/9 Herby Vojčík <[hidden email] <mailto:[hidden email]>>

   Hi, I'd like to propose two alternatives to inline JS and variable
   naming (they are not entirely unrelated):

   Proposal 1.
     - Let inline JS statement occur only in context of Smalltalk
   statements (as a member of "method statement sequence" or "block
   statement sequence"). In particular, it should be not allowed inside
   expressions, and
     - let names of variables (args, temps, block args, block temps,
   unknowns/globals) be unique over the whole text of the method (none
   of them can produce name clash)

   This goes with GH-18 issue and can help solve GH-191 issue as well.
   Plus, it helps making more optimized inlining in compiler.

   Proposal 2.
     - Let inline JS statement occur only in context of Smalltalk
   statement in a main method code (as a member of "method staement
   sequence"). In particular, it should not be allowed inside blocks
   not expressions, and
     - deny names of global/unknown variables which conflict with
   "smalltalk" variables (args, temps, block args, block temps).

   This may also help in the same issues, though a bit differently.

   Is one of these acceptable for amber?

   Herby

nicolas petton wrote:

Yeah, I meant the first one. Now what do you mean by unique over the
whole text of the method?

Just the "does not clash with any other type of the variable having the same name (including unknowns)".

Herby

Nico

2012/5/10 Herby Vojčík <[hidden email] <mailto:[hidden email]>>

   nicolas petton wrote:

       I fully agree :) A few changes in the parser and compiler should
       do the
       trick.


   Which one? I like the first, it's cleaner.

   Herby

       Nico

       2012/5/9 Herby Vojčík <[hidden email]

       <mailto:[hidden email]> <mailto:[hidden email]

       <mailto:[hidden email]>>>


           Hi, I'd like to propose two alternatives to inline JS and
       variable
           naming (they are not entirely unrelated):

           Proposal 1.
             - Let inline JS statement occur only in context of Smalltalk
           statements (as a member of "method statement sequence" or "block
           statement sequence"). In particular, it should be not
       allowed inside
           expressions, and
             - let names of variables (args, temps, block args, block
       temps,
           unknowns/globals) be unique over the whole text of the
       method (none
           of them can produce name clash)

           This goes with GH-18 issue and can help solve GH-191 issue
       as well.
           Plus, it helps making more optimized inlining in compiler.

           Proposal 2.
             - Let inline JS statement occur only in context of Smalltalk
           statement in a main method code (as a member of "method staement
           sequence"). In particular, it should not be allowed inside
       blocks
           not expressions, and
             - deny names of global/unknown variables which conflict with
       "smalltalk" variables (args, temps, block args, block temps).

           This may also help in the same issues, though a bit differently.

           Is one of these acceptable for amber?

           Herby