Add generic support for destructuring assignment in @each

[xzyfer]

This PR adds generic support for destructuring assignment in @each.

Fixes #492, #577. Specs added sass/sass-spec#54, sass/sass-spec#98, sass/sass-spec#99.

[xzyfer]

There's been a bit of activity tonight. I'm going to let this rest until tomorrow.

[coveralls]

Coverage decreased (-0.06%) when pulling 60a2107 on xzyfer:feat/each-maps-for-realz into 02f9058 on sass:master.

[am11]

Thanks for the fix. 👍

Can you please also make these changes (to stop msvcr from barfing: warning C4101: 'bc' : unreferenced local variable):

diff --git a/ast.hpp b/ast.hpp
index 59b9a4e..3bfac33 100644
--- a/ast.hpp
+++ b/ast.hpp
@@ -658,7 +658,7 @@ namespace Sass {
           if (!(*(elements()[i]) == *(l[i]))) return false;
         return true;
       }
-      catch (std::bad_cast& bc)
+      catch (std::bad_cast&)
       {
         return false;
       }
@@ -706,7 +706,7 @@ namespace Sass {
           if (!(*at(key) == *m.at(key))) return false;
         return true;
       }
-      catch (std::bad_cast& bc)
+      catch (std::bad_cast&)
       {
         return false;
       }
@@ -815,7 +815,7 @@ namespace Sass {
         Variable& e = dynamic_cast<Variable&>(rhs);
         return e && name() == e.name();
       }
-      catch (std::bad_cast& bc)
+      catch (std::bad_cast&)
       {
         return false;
       }
@@ -969,7 +969,7 @@ namespace Sass {
         e.normalize(find_convertible_unit());
         return unit() == e.unit() && value() == e.value();
       }
-      catch (std::bad_cast& bc)
+      catch (std::bad_cast&)
       {
         return false;
       }
@@ -1008,7 +1008,7 @@ namespace Sass {
         Color& c = (dynamic_cast<Color&>(rhs));
         return c && r() == c.r() && g() == c.g() && b() == c.b() && a() == c.a();
       }
-      catch (std::bad_cast& bc)
+      catch (std::bad_cast&)
       {
         return false;
       }
@@ -1044,7 +1044,7 @@ namespace Sass {
         Boolean& e = dynamic_cast<Boolean&>(rhs);
         return e && value() == e.value();
       }
-      catch (std::bad_cast& bc)
+      catch (std::bad_cast&)
       {
         return false;
       }
@@ -1120,7 +1120,7 @@ namespace Sass {
         String_Constant& e = dynamic_cast<String_Constant&>(rhs);
         return e && unquoted_ == e.unquoted_;
       }
-      catch (std::bad_cast& bc)
+      catch (std::bad_cast&)
       {
         return false;
       }

Also, it gives few other warnings: https://gist.github.com/am11/5324b8ffb44f0a599329. It might make sense to fix them, perhaps to save couple of milliseconds here and there? #519 is related.

[mgreter]

Can we also get rid of the size_t hash_ = 0 initialization in ast.hpp? It breaks my travis-ci builds for perl-libsass (I'm compiling with -std=c++0x). Don't know where to initialize them properly though.

[am11]

Guys I have made the changes. Please review #572.

[QuLogic]

Can we also get rid of the size_t hash_ = 0 initialization in ast.hpp? It breaks my travis-ci builds for perl-libsass (I'm compiling with -std=c++0x). Don't know where to initialize them properly though.

Probably have to move it to the initialization list in the constructor (after the name and : but before the {).

[mirisuzanne]

@xzyfer This branch passes all my basic map tests. I ran this series of tests (with the final @each loop un-commented):

.test-each {
  -map-get: 'yep';
  -map-remove: 'yep';
  -map-merge: 'yep';
  -map-keys: 'yep';
  -map-values: 'yep';
  -map-has-key: 'yep';
  -map-compare: 'yep'; }

That seems to fix merge, compare, and loop. I'll also run it against larger frameworks when I get a chance. Cheers! :)

[xzyfer]

I'd prefer to limit the scope of this PR. Focused PRs make for better auditing and regression tracking.

@am11 thanks for opening that PR. Looks fine from a cursory glance. I'll take a closer look after work.

@mgreter @QuLogic I'll look at how I can move the hash_ initialisation. It's used for most Expressions not just lists, but is tied into Vectorized and Hashed. My feeling is that there will need to be a better interface to handle hash caching. Again I'd prefer to handle this in another PR since it's related to different feature.

[xzyfer]

@akhleung I'm a bit foggy on the purpose inspect.ccp, especially in regards to control structures like @each. Could you please shed some light?

[akhleung]

@xzyfer The Inspect visitor is meant for rendering internal nodes back into a textual format (and the To_String visitor is defined using Inspect). In addition to rendering the final CSS output, it's also useful for debugging, in case you just need to print out internal nodes in your debug logs.

Control directives such as @each aren't present in the fully expanded tree, so they'll never be rendered in the final output. But the Inspect visitor still handles them in case you really need to print them out.

[xzyfer]

Control directives such as @each aren't present in the fully expanded tree, so they'll never be rendered in the final output. But the Inspect visitor still handles them in case you really need to print them out.

Ah great this is the use case I wasn't aware of. Thanks for the clarification.

[xzyfer]

Thanks @ericam ! The map api should work as expected on the current master as of about 12hrs ago 🎉

Any edge cases you stumble across in @each would be great to test.

[mgreter]

The tip by @QuLogic gave me the right directions, please see here for a IMO proper fix: mgreter@4fa3014

[xzyfer]

Looks sane to me! Can you put that in a PR?

[xzyfer]

This now passes all the specs. I'll wait for the above spec to be merged before merging this.

[coveralls]

Coverage decreased (-0.1%) when pulling 9cac507 on xzyfer:feat/each-maps-for-realz into 02f9058 on sass:master.

[coveralls]

Coverage decreased (-0.11%) when pulling 0a643df on xzyfer:feat/each-maps-for-realz into 52ad379 on sass:master.

[coveralls]

Coverage increased (+0.09%) when pulling e8c549a on xzyfer:feat/each-maps-for-realz into 52ad379 on sass:master.

[HamptonMakes]

If I'm reading the above thread correctly, seems like this is ready to be merged in. So, I'm doing it... like now. ;P