Re: weird macro params issue

Jonathan Revusky wrote:
> On Fri, May 2, 2008 at 7:18 PM, Attila Szegedi <[email protected]> wrote:
>> The order of populating defaulted variables is undefined -- they're
>>  being pulled out by an iterator out of a hash map. So if this worked
>>  for you, then you just had blind luck so far, I'm afraid.
>>
>>  We could ensure the order going forward, or -- since with named-
>>  argument invocations there's not really any specific order -- even
>>  code it so that it fills out default expressions iteratively: go over
>>  them once, resolving what it can; then repeat until it gets an
>>  iteration where either (a) all default expressions got resolved or (b)
>>  it got something other than an InvalidReferenceExcepion or (c) it
>>  could not resolve any additional default expressions (and in this
>>  case, rethrow the InvalidReferenceException for one arbitrarily chosen
>>  unresolved default expression). That'd support any weird defaulting
>>  dependency order as long as there are no circular references, and
>>  circular references would also fall out with an
>>  InvalidReferenceException.
>>
>>  Anyone object to having this logic in a subsequent FreeMarker release?
> 
> Just offhand, I wouldn't encourage anybody to do something like that.
> It surprised me that it worked at all, actually. I would have to look
> at the implementation. It seems to me that the default parameters
> should be evaluated in the scope of the enclosing namespace and thus,
> what Ioannis was doing should fail.
> 
> Now, of course, the current situation, that it works sometimes and
> other times doesn't work, that's extremely undesirable. But to be
> honest, it seems like such an overly intricate, excessively clever way
> of doing things, that I would prefer for it just to always fail. :-)
> 
> I mean, seriously, is this coding pattern actually useful?
> 
> Of course, it would be better if it always worked than if it always
> failed, 

I misspoke above. I meant, here, that it would be better if it always 
worked than if it worked some of the time and failed other times. I 
mean, it should always work or always fail... :-)

JR
but I think I would just tend towards it always failing.
> 
> JR
> 
> 
> 
>>
>>  Attila.
>>
>>  On 2008.05.02., at 10:58, John Cherouvim wrote:
>>
>>
>>
>>> Hello
>>  >
>>  > I have this macro:
>>  > <#macro test a b=a c=b>
>>  >    ${a} - ${b} - ${c}
>>  > </#macro>
>>  >
>>  > And this client code:
>>  > <@test 1 /><br/>
>>  > <@test 1 2 /><br/>
>>  > <@test 1 2 3 /><br/>
>>  >
>>  > Which produces:
>>  > 1 - 1 - 1
>>  > 1 - 2 - 2
>>  > 1 - 2 - 3
>>  >
>>  > Now if I just change the param names of the macro:
>>  > <#macro test foo bar=foo baz=bar>
>>  >    ${foo} - ${bar} - ${baz}
>>  > </#macro>
>>  >
>>  > And keep the same client code, then the output produced is:
>>  > InvalidReferenceException: Expression bar is undefined on line 20,
>>  > column 31 in index.ftl.
>>  > 1 - 2 - 2
>>  > 1 - 2 - 3
>>  >
>>  > The first call fails when it tries to evaluate the baz=bar. Why is
>>  > that?
>>  >
>>  > Also, is this usage valid in freemarker? To have parameters default to
>>  > the value of other parameters. I find it very useful for some use
>>  > cases.
>>  >
>>  > thanks,
>>  > Ioannis
>>
>>
>>  -------------------------------------------------------------------------
>>  This SF.net email is sponsored by the 2008 JavaOne(SM) Conference
>>  Don't miss this year's exciting event. There's still time to save $100.
>>  Use priority code J8TL2D2.
>>  http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone
>>  _______________________________________________
>>  FreeMarker-user mailing list
>>  [email protected]
>>  https://lists.sourceforge.net/lists/listinfo/freemarker-user
>>
> 


-------------------------------------------------------------------------
This SF.net email is sponsored by the 2008 JavaOne(SM) Conference 
Don't miss this year's exciting event. There's still time to save $100. 
Use priority code J8TL2D2. 
http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone