Re: [Gaea] Named optional attributes

On 11/6/2011 9:08 AM, Joel de Guzman wrote:
> On 11/6/2011 5:58 AM, Matias Capeletto wrote:
>>>> [var footnotes-list []]
>>>>  [set footnotes-list [append [footnotes-list] [...]]
>>>
>>> TBH, I'm not quite happy with it either. The language is purely
>>> functional apart from that. There should be a way to do above using
>>> a purely FP way, but it's not as easy.
>>
>> At least it follows scheme's (define x y) (set! x z).
>>
>>> As always, suggestions welcome.
>>
>> I was asking about other uses cases, because if it is only introduced
>> to deal with footnotes in HTML maybe they shouldn't need to be
>> introduced right now if it is only used to produce LaTeX or Docbook.
>> That way you can get some time to see how the real implementation
>> works.
> 
> Indeed.
> 
>> If they are introduced, I am tempted to add a [push mutable-var
>> element] construct so the collectors are easier to write.
> 
> Well, that's easy:
> 
>   [def [push x e]
>     [set [x][append [x][e]]]
>   ]
> 
> What's interesting is that you should be able to pass 'var'
> arguments much like C++ references.
> 
> The side-effects make me feel uneasy though. Here, push
> has side-effects while append (and the rest of the gang)
> does not :-(

If we go this route (still unsure), perhaps we can make it clear
by having a naming protocol that makes templates with side-effects
clear. E.g. we can have in-place versions for:

  [decl [append! x e]]
  [decl [insert! x e n]]
  [decl [reverse! x]]
  [decl [join! x y]]

Even:

  [decl [transform! x f]]

This will open up a whole slew of templates like:

  [decl [for-each! x f]]

Perhaps we should then rename [set x v] --> [set! x v]

Is this good or bad? I don't know. Well, scheme and lisp do
have side-effects. But, I kinda agree with Terrence Parr:
"The fact that StringTemplate does not allow such things
as assignments (no side-effects) should make you suspicious
of engines that do allow it."

Regards,
-- 
Joel de Guzman
http://www.boostpro.com
http://boost-spirit.com