Grzegorz Kossakowski pisze: > Reinhard Poetz pisze: >> It's time to release Cocoon 2.2, finally! We have been working on it for >> years and I think it's time to ship the *final* release. I want to do >> this in three phases: >> >> 1) During the first phase I will release our two sub-projects "Cocoon >> Configuration" and "Cocoon Servlet-Service-Framework". This time I will >> create the Maven 2 release artifacts and "normal" zip/tar release >> artifacts for non-Maven users. >> >> I think that I will be able to finish it by the end of next week. >> > > There is a subtle problem I discovered few days ago but I haven't reported up > to date because I > hadn't enough free time to confirm if this problem really exist. Anyway, I > think that > cocoon-servlet-service-impl is not really Cocoon-independent because of this > line: > > resolver = (SourceResolver) factory.getBean(SourceResolver.ROLE); > > of getResource() method from ServletServiceContext class. The problem I can > see here is that if > cocoon-servlet-service-impl is used outside the Cocoon there is no default > implementation of > SourceResolver registered as a Spring bean. It's not a big deal, because > Excalibur provides such > implementation but we would need to figure out how to register and properly > set up this component. > > Actually, I see whole getResource() method flawed because it calculates > contextPath when used first > time and I think this code should belong to the setup (not runtime) phase of > ServletServiceContext. > I already tried to move this code (I paste the patch at the end of this > e-mail) but stumbled across > some NPEs for specific servlet bean configurations and I haven't had enough > free time to dig into > this. I planned to take care of it during upcoming weekend because I finally > get done with all these > exams at my university. > > The root motivation for me taking a look at this code was some > experimentation with Micro Cocoon. > Specifically, I tried to make CocoonSourceResolver casual Spring bean and due > to flawed contextPath > initialization I ran into cyclic dependency chain. > > To be honest, I don't have an idea if we should consider this a blocker. I > lean towards not > considering this a blocker but once we sort this out we will need to release > 1.1.0 version of > cocoon-servlet-service-impl. If I come up with the fix until Monday, do you > think it's a good idea > to commit it just before final release? (I think it shouldn't harm and I > would give it very > throughout testing)
Since I got no feedback I decided to commit my changes. I gave it quite detailed testing and I don't think it may break anything. Anyway, if you have spare cycles, it would be great if you could test if everything still works ok. -- Grzegorz Kossakowski
