Thinking it over again, and looking at the 'spirit' of the current coding guideline, I'm not so sure it would fit well. The current coding guideline is on the _layout_ of the code, not on the meaning of it (where the usage of the final keyword really conveys something about your intentions). Isn't this something we should consider to be common practice?
Any other opinions on the list? Angelo On Nov 17, 2010, at 8:35 PM, Angelo van der Sijpt wrote: > Hi, > > On Nov 17, 2010, at 3:26 PM, Bram de Kruijff wrote: > >> On Wed, Nov 17, 2010 at 12:02 PM, Angelo van der Sijpt >> <angelo.vandersijpt at luminis.eu> wrote: >>> Hi, >>> >>> Good to see we have a file-based storage for our tenants now! >>> >>> Still, there are some things I don't really understand. >>> - I noticed the storage provider can be configured using a system property; >>> why not use the ConfigurationAdmin for this? >> >> Good point. I was writing this under the assumption I might have to >> stick it under ConfigAdmin itself, but I see the Felix ConfigiAdmin >> has its own filebased storage allready (and uses a Java Property to >> allow overriding bundle storage). I'll change this. > > Okidoki. > >> >>> - The FSTSP synchronizes all its methods; this it not necessary, as the >>> TenantStorageProvider javadoc states 'The TenantStorageProvider can assume >>> that synchronization will be handled externally.' >> >> I see... but I dont feel comfortable not synchronizing when I know my >> impl of a published service cannot handle it. On the otherhand this is >> an SPI so maybe I'm being a little too defensive :) > > In essence I'm all for defensive programming, but as you have noted, this is > an SPI with a note on concurrency. By the way, we should come up with a > standard pattern for these situations. > >> >>> - It seems you use finally a little differently than I do (e.g., in >>> FSTentantIdList, the m_file is final, but the m_tenantIdList is not). Our >>> coding guideline does not state anything about this; should we include >>> that? Something like "members should be made final if they can, and local >>> variables should only be made final when they have to be". >> >> Seems I forgot to put the final modifier on m_tenantIdList. In general >> I like making things final where possible (and sometimes relevant). No >> strong opinion with regard to local variables though so +1 for your >> proposal :) > > I will put a proposal up on the wiki. > >> >> Regards, >> Bram > > Angelo > >> >> _______________________________________________ >> Amdatu-developers mailing list >> Amdatu-developers at amdatu.org >> http://lists.amdatu.org/mailman/listinfo/amdatu-developers >
