Re: Varargs

Ah… okay, I totally forgot about that. Alright, that means that what you suggested as the fix should be sufficient.

On Aug 8, 2011, at 6:35 PM, Daniel Dekany wrote:

> Just so it's absolutely clear: The vararg support was already
> backported from 3.0 to 2.3.12, almost 4 years ago. But this
> backporting was wrong (and obviously it wasn't even tried if it
> works). So this should be fixed in 2.3 too, not just in 2.4.
> 
> -- 
> Best regards,
> Daniel Dekany
> 
> 
> Monday, August 8, 2011, 5:50:51 PM, Attila Szegedi wrote:
> 
>> Yeah, I think that would be sufficient. I did implement full vararg
>> support for 3.0, so whatever code is there should work. 
>> 
>> There's some additional work to be done at invocation time; i.e. if
>> the last argument is already an array of the requested type, you
>> should just pass it, otherwise you should construct an array. Also,
>> if one less argument is passed, you need to pass in a zero-length array.
>> 
>> Finally, there's a whole subset of issues with overloaded method
>> resolution in presence of varargs. Again, I did write all the code
>> for 3.0, and it should be straightforward to port it back to 2.4, we
>> just need someone to roll the sleeves up and do it. That's not me at the moment, unfortunately.
>> 
>> If Brian forks the project on GitHub, that's most welcome. We'll be happy to pull in changes.
>> 
>> Attila.
>> 
>> On Aug 4, 2011, at 11:46 AM, Daniel Dekany wrote:
>> 
>>> OK, I have looked into this, and it seems that this feature (varargs)
>>> has never worked in 2.3. FM 2.3.12 has just made things worse, as not
>>> even explicitly passing an array (like hasRoles(['foo', 'bar'])) works
>>> starting from there, while before 2.3.12 it certainly did. How nobody
>>> ran into this and reported for 4.5 years is beyond be. Also, varags
>>> has no test coverage in 2.3/2.4...
>>> 
>>> And now I will mostly talk to Attila (the author of BeansWrapper). The
>>> problem is this: BeansWrapper.populateClassMapWithBeanInfo(Class) in
>>> FM 3.0 calls componentizeLastArg(publicMethod.getParameterTypes(),
>>> publicMethod.isVarArgs())) to calculate SimpleMemberModel.argTypes,
>>> while 2.3/2.4 just calls publicMethod.getParameterTypes(). The
>>> componentizeLastArg method changes the vararg's type from array<T> to
>>> T. Thus in 2.3/2.4 the type of the vararg argument, according to
>>> SimpleMemberModel.argTypes[theIndexOfTheVarArg], will be array<T>,
>>> while in 3.0 it will be T. But all the other parts of the algorithm is
>>> the same as in 3.0, so the method call will fail both with
>>> hasRoles(["foo", "bar"]) and hasRoles("foo", "bar").
>>> 
>>> Would adding componentizeLastArg to 2.3/2.4 and calling it at the same
>>> two places as in 3.0 be a sufficient fix? Would it be backward
>>> compatible? I guess it would be, but you are still the one who knows
>>> this stuff inside out...
>>> 
>>> -- 
>>> Best regards,
>>> Daniel Dekany
>>> 
>>> 
>>> Wednesday, August 3, 2011, 6:19:59 PM, Brian Pontarelli wrote:
>>> 
>>>> On Aug 3, 2011, at 2:07 AM, Daniel Dekany wrote:
>>>> 
>>>>> Tuesday, August 2, 2011, 9:20:43 PM, Brian Pontarelli wrote:
>>>>> 
>>>>>> I've read that vararg method calls were added to 2.4.12 but I can't
>>>>>> find any good docs on how to use them and if there are any special
>>>>>> requirements. I have a varargs method that I'm trying to invoke from
>>>>>> a template and I'm getting an argument type mismatch exception:
>>>>>> 
>>>>>> public boolean hasRoles(String... roles) {
>>>>>>  ...
>>>>>> }
>>>>>> 
>>>>>> 
>>>>>> [#if user.hasRoles(["foo", "bar"])]
>>>>>>  …
>>>>>> [/#if]
>>>>>> 
>>>>>> Anyone know how to fix this?
>>>>> 
>>>>> AFAR it's supposed to Just Work (assuming you are using some
>>>>> BeansWrapper subclass as the object_wrapper setting, and it's 99% that
>>>>> you do). Have you tried user.hasRoles("foo", "bar"), without the []?
>>>> 
>>>> Yeah, tried that. It then complains about trying to convert String
>>>> to String[]. I do have a BeansWrapper sub-class, but all it does is set these two options:
>>>> 
>>>>       setExposeFields(true);
>>>>       setSimpleMapWrapper(true);
>>>> 
>>>> Not sure how to fix it, but I changed the method to take an
>>>> Iterable<String> and it is working.
>>>> 
>>>> -bp
>>>> 
>>>> 
>>>> ------------------------------------------------------------------------------
>>>> BlackBerry&reg; DevCon Americas, Oct. 18-20, San Francisco, CA
>>>> The must-attend event for mobile developers. Connect with experts. 
>>>> Get tools for creating Super Apps. See the latest technologies.
>>>> Sessions, hands-on labs, demos & much more. Register early & save!
>>>> http://p.sf.net/sfu/rim-blackberry-1
>>>> _______________________________________________
>>>> FreeMarker-user mailing list
>>>> [email protected]
>>>> https://lists.sourceforge.net/lists/listinfo/freemarker-user
>>>> 
>>> 
>> 
>> 
>> ------------------------------------------------------------------------------
>> BlackBerry&reg; DevCon Americas, Oct. 18-20, San Francisco, CA
>> The must-attend event for mobile developers. Connect with experts. 
>> Get tools for creating Super Apps. See the latest technologies.
>> Sessions, hands-on labs, demos & much more. Register early & save!
>> http://p.sf.net/sfu/rim-blackberry-1
>> _______________________________________________
>> FreeMarker-user mailing list
>> [email protected]
>> https://lists.sourceforge.net/lists/listinfo/freemarker-user
>> 
> 


------------------------------------------------------------------------------
BlackBerry&reg; DevCon Americas, Oct. 18-20, San Francisco, CA
The must-attend event for mobile developers. Connect with experts. 
Get tools for creating Super Apps. See the latest technologies.
Sessions, hands-on labs, demos & much more. Register early & save!
http://p.sf.net/sfu/rim-blackberry-1