Re: [morph-developer] recent changes

[Matt S. <Mat...@wh...>]

Wow I'm slow getting back to you ><  All these changes look great!  I'm 
very glad you have SVN access for this stuff :)

Here are some minor changes I made:
- ReflectionInfo class, took your change of !isLowerCase and further 
refined it to be isUpperCase.  Also applied the change in multiple 
places (you only caught 1 of the 3 places the check happened)
- NumberToBooleanConverter, added comment indicating .equals cannot be 
used (to prevent someone going in there and messing it up in the future)
- ObjectToClassConverter, added comment explaining why 
isTransformableImpl was implemented (for performance)

Regarding the MultipleDestinationConverter, the final call in the class 
was probably supposed to be 
getContainerConverter().convert(destinationClass, *destinationObjects*, 
locale);  I went ahead and made this change.  This class was created 
because someone was challenging me that my interfaces only allowed 1 -> 
1 object conversions as opposed to also allowing one object to be broken 
into several smaller object or several smaller objects to be combined 
into 1 larger object.  The idea here would be you had something like, 
say, a Person that you wanted to be converted into a collection of 
smaller objects like PoliticalViews, BankAccount, Cars.  Those 3 smaller 
objects would have to be returned as some type of collection, say as a 
List or an array of Object[]s.  That was what the final call to 
getContainerConverter() was for.  Really this is more of a hypothetical 
class to show that this type of thing was possible to my friend I was 
hoping would get involved in the project (didn't work).  I don't 
anticipate anyone would actually use this class, and I probably haven't 
even looked at it since the date it was first written.  Perhaps we 
should just delete it?

Question: why did you put in clone() calls in the TimeConverter, if the 
source class = destination class?  By way of compmarison, the 
IdentityConverter does not attempt to make copies of source objects and 
just returns them unchanged.  The tradeoff as I see it is that clone() 
is "safer" but just returning the source unchanged is faster.  Since we 
can't universally clone objects (since not all objects implement that 
interface), would it be best to just never do cloning of source objects?

Suggestion: how about we rename Int to MutableInteger, to more 
accurately describe what it is for?  This would also preclude confusion 
regarding what is returned by the Morph.convertToInt method.

FYI: I made some improvements to MorphPropertyEditor, which I am 
actively using in one of my apps to customize Spring's data binding 
behavior (I happen to need some behavior that is not Morph's default, 
specifically related to the TextToNumberConverter)

Matt

Matt Benson wrote:
> Matt S (and anyone else who may be watching):
>
> I have been working in the Morph codebase yesterday
> and today.  Highlights of what I've done:
>
> r10 | matt.benson | 2007-01-08 13:38:04 -0600 (Mon, 08
> Jan 2007) | 3 lines
>
> Fix/improve date-related test failures by sticking to
> Calendar instead of Date
> whenever possible.
>
> -Allowed me to establish a baseline of all-tests-pass.
>
> r11 | matt.benson | 2007-01-08 13:49:02 -0600 (Mon, 08
> Jan 2007) | 2 lines
>
> Fix ChainedConverter so that it will work correctly
> for ExplicitTransformers.
>
> -The original bug I saw that I thought I could fix.
>
> r17 | matt.benson | 2007-01-08 16:38:16 -0600 (Mon, 08
> Jan 2007) | 1 line
>
> make text converter handle byte and char arrays
>
> -Seemed like the right thing to do.
>
> r18 | matt.benson | 2007-01-08 16:48:00 -0600 (Mon, 08
> Jan 2007) | 1 line
>
> delegate text to boolean conversion using a
> textconverter, thus gaining any additional source
> classes added to textconverter for free
>
> -followed naturally from the previous change.
>
> r19 | matt.benson | 2007-01-08 16:51:31 -0600 (Mon, 08
> Jan 2007) | 1 line
>
> fix broken stack management and add object to pretty
> text converter test case
>
> -ObjectToPrettyTextConverter had no test case and
> didn't work properly.  :(
>
> r29 | matt.benson | 2007-01-08 17:22:06 -0600 (Mon, 08
> Jan 2007) | 2 lines
>
> add missing primitives; quicker isTransformableImpl();
> eliminate code duplication
>
> -I couldn't think why ObjectToClassConverter allowed
> all primitives as sources except for boolean and char,
> so I added them.  Since this converter's only
> destination Class is Class, and Class is final, I
> overrode isTransformableImpl to check whether the
> destinationType were Class.class...
>
> r32 | matt.benson | 2007-01-08 17:30:32 -0600 (Mon, 08
> Jan 2007) | 1 line
>
> add UnsupportedOperationException to
> CombiningCopier.copyImpl()
>
> -not yet implemented
>
> Additionally I made several changes having to do with
> eliminating redundant (explicitly or logically) code,
> reducing method calls, object instantiations, source
> file size, etc. where I saw possibilities.  Some of
> these could be considered style issues so I hope I
> haven't overstepped any bounds here.
>
> br,
> Matt B
>
> __________________________________________________
> Do You Yahoo!?
> Tired of spam?  Yahoo! Mail has the best spam protection around 
> http://mail.yahoo.com 
>
> -------------------------------------------------------------------------
> Take Surveys. Earn Cash. Influence the Future of IT
> Join SourceForge.net's Techsay panel and you'll get the chance to share your
> opinions on IT & business topics through brief surveys - and earn cash
> http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
> _______________________________________________
> morph-developer mailing list
> mor...@li...
> https://lists.sourceforge.net/lists/listinfo/morph-developer
>
>