Geoff Howard [mailto:[EMAIL PROTECTED] wrote: > On 12/20/05, Max Pfingsthorn <[EMAIL PROTECTED]> wrote: > > > > On another note: I need the eventcaching block for webdav, but > > > > that one only needs jms in one class, and databases in the > > > > samples. So, I'll work on the dependency issue there instead of > > > > in the webdav block directly. > > ... > > > The eventcache is needed for more advanced caching. The > components need to know about it to be able to construct the > right Validity objects for Source.getValidity(). We found out > that eventcaching is really key for good performance of the > website, so I would consider it a good kind of dependency. Of > course, the eventcaching block depending (indirectly) on the > database block is a bit silly. > > Yes, these dependencies were always somewhat painful - as we discussed > before [1]. It's only the samples that cause the dependency on the > database block IIRC. There was some work being done on samples > dependencies I think - or were samples being separated into samples > blocks perhaps? That would cure this. > > I see you've implemented some of this in webdav - did you manage to > avoid a dependency on the database block somehow?
Yes, well, at least directly. The webdav block now depends only on repository and eventcache, not on database. However, eventcache still depends on database. I was thinking about the same thing, meaning to make a new block for the eventcache samples. That has been done for other blocks and would take care it, as you said. However, I don't know if its worth it in the 2.1 branch. Compiling a few more classes doesn't hurt too much for now. It would make more sense and be worth it for 2.2 as it is supposed to be released semi-soon, right? max > [1] http://marc.theaimsgroup.com/?t=113259286500003&r=1&w=2