Re: [Springframework-developer] Input on JIRA issue

[Glen M. <gr...@ve...>]

I would advise against *complete* backwards compatibility, especially if 
it means burdening future releases of Spring with seldom-used cobwebs 
from previous ones, of things that originally sounded like good ideas 
but turned out not to work so well after all.  I think it is important 
that Spring have some "nimbleness", i.e. occasionally be allowed to 
clear itself out of past baggage, so it can compete effectively against 
future competitor frameworks, that, by virtue of being new, don't need 
to be burdened with older API's from past releases.

Glen

Colin Sampaleanu write:
> I don't think this will affect many people at all, as it's quite rare 
> for people to use TransactionProxyFactoryBean programmatically. You 
> could almost break things down into backwards compatibility for the 
> declarative use (very very common) vs. programmatic use.
> 
> I definitely like the use of Class[] more, it's cleaner. However if we 
> are really concerned about even programmatic backwards compatibility, 
> then one consideration is that in 2.0 people won't even be using 
> TransactionProxyFactoryBean very much (preferring the new XML dialect), 
> so it's not like the "benefit" of the Class[] signature is getting to 
> many people.
> 
> Of course, the 2.0 class could support both signatures, but the in that 
> case for the declarative use the String[] will always be picked for the 
> declarative case as it's a one-to-one mapping.
> 
> Colin
> 
> On 7/12/2006 10:15 AM, Matt Sgarlata wrote:
> 
>>My 2 cents is that an IDE will point this out to you and the change will 
>>take all of 2 minutes to code.
>>
>>It would be a bigger deal IMO if you had to change your context 
>>definitions, because in that case the error might not be found for a 
>>while (for example, if the bean was lazily initialized).  However, 
>>Spring's use of PropertyEditors means the change should be transparent 
>>in terms of XML application context definitions.
>>
>>Matt
>>
>>Rob Harrop wrote:
>>  
>>
>>>All,
>>>
>>>I would like your input on this JIRA issue. Do you think this is valid 
>>>breakage of backward-compatibility or would you like to see the change 
>>>backed out?
>>>
>>>http://opensource.atlassian.com/projects/spring/browse/SPR-2277
>>>
>>>Rob
>>>