I was looking at the adapter classes in solr and was wondering if we should refactor these and had the idea that these could be just subclasses of AnyMap.
Anyhow, i solved the problem otherwise and its not pressing but was just wondering why they are final. I tend to not make anything final as programmers should know what they are doing and limiting often hinders elegant solutions or short time hacks / at least I have been stumped a few times due to a final and had to work around it.
In this case we have already the interfaces protecting normal users from thinking, “oh let’s just subclass it”, so I think there is no real danger in allowing this.
Thomas Menzel @ brox IT-Solutions GmbH
From: smila-dev-bounces@xxxxxxxxxxx [mailto:smila-dev-bounces@xxxxxxxxxxx] On Behalf Of Jürgen Schumacher
Sent: Freitag, 20. Januar 2012 15:25
To: Smila project developer mailing list
Subject: Re: [smila-dev] datamodel impls are final
Hi,
I never had this problem or the need to subclass these classes. What do you want to achieve with it? Could you give an outline of such a test case?
In general, I don’t think that subclassing should be the first choice of achieving thinks … most of the time it causes more problems than it seems to solve.
Cheers,
Juergen.
Hi folks,
What is the reason for the impl classes being final?
I find it a bit cumbersome. Shure nobody should really subclass this for real but sometimes this is convenient, particularly for tests.
I vote to just doc that subclassing is at own risk, after all they are in the impl. package
Thomas Menzel @ brox IT-Solutions GmbH