To repeat what Remy has been saying, if you think you need to know how the plugin system works you are Doing It Wrong (tm). That isn't to say there aren't a few times here and there where that is the only option, but you should probably leave those alone for the most part. In this particular case you should make each of the models take the env as an argument, so that you can pass it to the component to get an instance.
--Noah On Oct 28, 2008, at 9:36 AM, anatoly techtonik wrote: > > On Oct 28, 4:48 pm, Remy Blank <[EMAIL PROTECTED]> wrote: > >>> If ComponentManager >>> implements singleton pattern then it doesn't matter where it is >>> instantiated. >> >> But Components are *not* singletons. There is exactly one instance of >> each Component *per ComponentManager*, so in the case of Trac, *per >> environment*. > > Got it. Thanks! I've added a minor comment to wiki that component > managers are not singletons. However, it seems that the role of > Component Manager deserves a separate paragraph. > >>> If I >>> understand correctly the only outcome of skipping the Environment in >>> my case is that my extension point can not be disabled in trac.ini >> >> No, it means that when you call CacheManager(ComponentManager()), you >> will instantiate a new manager and a new component every time. > > The strange thing that in spite of CacheManager isolated in separate > ComponentManager, the extension point is registered correctly and > plugin is also called when needed. If I understand it right - > implementing plugins should be registered by the Environment in the > Environment. But how does it happen that they are registered as > isolated CacheManager observers? > >> There's another complexity that is hidden. When you run Trac in >> Apache >> with worker threads and mod_python, and multiple projects, you will >> have >> several environment loaded *at the same time*, one for each project. >> Every environment will instantiate every active component once. The >> environment is what ties these instances together. > > I see that my messages in trac.log are duplicated even without > mod_python with FastCGI. Does that mean they are using some lock > mechanism to access shared resources? Should plugins be aware and use > some API to do the same? I am using MySQL, but it would probably > create problems with SQLite if I decide to make my own database in > trac environment directory for whatever reason I have. Is it right? > >> But really, this is all plumbing that you shouldn't worry about. Your >> only problem is that you don't have access to the environment from >> inside CachedRepository, so you have to find a way to get it there. >> That's all. > > No-no-no. The whole thing is very interesting and it is not described > anywhere. > >>> Inherited and initialized. The line 50 is located inside of >>> observer's >>> method. >>> http://python.pastebin.com/d2f6044ae >> >> That looks correct indeed. It is most certainly related to your >> instantiation of a ComponentManager. Fix the problem above, and the >> rest >> should start working. > > Sorry for taking your time, but I won't be able to sleep if I coudn't > get why my plugin observer does not receive required environment > attributes if extension point component was instantiated from other > component manager? I will try to investigate the code further > tomorrow. > > Thanks for support. =) > -- > WBR, > anatoly > > > --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "Trac Development" group. To post to this group, send email to [email protected] To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/trac-dev?hl=en -~----------~----~----~----~------~----~------~--~---
