issue with variable scope in namespaces

[nlac]

Hello,
please check the following code compiled flawlessly by less.js:

/* mixin library uses a variable from the "caller" namespace #skin */
#lib {
  //@w : @size;
  .rect() {
    width:2*@size;
    height:@size;
  }
}

/* "caller" namespace uses the mixins from #lib in its mixins */
#skin {
 @size : 100px;
  .panel() {
    #lib > .rect();
  }
}

/* constructing actual html class */
div.intro {
  #skin > .panel();
}

The issue: if i remove the comment before the @w : @size; the compilation fails.

Is there a particular reason why? Why the mixin body can see the @size variable and the variable defining rule can't, in the #lib namespace?

Regards, nlac

[nlac]

Hi,
#lib is a namespace (go to http://lesscss.org and study the section "Namespaces"), just like the #skin, and it is intended to be a namespace - #lib may contain other rules. I want that rules to see variables, classes, mixins from #skin, that's what i'm doing.

[nlac]

which part shouldn't work? which part doesn't work in the correct way? how is it connected to the theory that "namespaces are still rules"? How your comments lead closer or even connected to the solution or at least the explanation of the issue?

[nlac]

My problem is, this one is the 7th comment and i (and people who also interest how scopes work in deep) didn't get any closer to the solution.

The original scope issue exposed in this ticket as a reminder, is:

why the compilation fails if i remove the comment in the 2nd line, while the .rect() mixin can flawlessly see the @size variable in the actual context?
It is NOT about how someone define a namespace in his head or how not.

I'm waiting for some people who can see the problem exposed.

[souldreamer]

It's because #lib doesn't know what @size is. #lib > .rect()'s code is pasted into #skin and tied to the @size var in #skin (because that's where the mix-in is instantiated), but @w isn't (as it has no ties to @size, and can't be computed so the parser has no idea what to do with it). Then #panel pastes the code for #skin > .panel() which is already tied to #skin's @size variable (i.e. setting @size in div.intro will do nothing).

[nlac]

Thanks souldreamer, that's clear and makes sense.

actually i wanted to customize semantic.gs a bit to be able to define more than one grid system (embedding the gs lib into a namespace with all its vars and mixins, @gridsystem-width would be the @w in the purified example etc.).

Thinking over your explanation i understand that the above example is not an issue and have to work around some other way. I close the issue.