RE: Performance (From the Mach-II Group) Kurt Wiersma said the following on 6/7/2007 9:06 AM: > Are you using ColdSpring's autowiring for your model or all the > relationships defined in your CS configuration file(s)? Autowiring is > very slow by nature of introspecting all the CFCs. Also, Chris Scott > mentioned to me the other day that CS is up to five times slower when > using deeply nested constructor-args via plain old property I think the "performance hit" was reloading the M2 framework during development, so performance probably was the wrong word. I haven't noticed any significant slowness, but my config files aren't all that big right now. I suspect M2 in production (-1) would only see any slowness the first time the app was loaded. Autowiring is the right solution for me. I didn't realize that without the autowiring I would have to explicitly define the properties in the CS config file, which I was trying to avoid since it seemed to defeat one of the purposes I was trying to achieve with inheritance. All in all a good learning experience - I now understand CS even better and am trying to figure out how I managed all my CFCs this long without it. anthony
________________________________ From: [email protected] [mailto:[EMAIL PROTECTED] On Behalf Of Brian Kotek Sent: Tuesday, June 12, 2007 6:23 AM To: [email protected] Subject: [coldspring-dev] Config/Inheritance Question On 6/11/07, Anthony Israel-Davis <[EMAIL PROTECTED] > wrote: I had turned global autowiring off because I just read on the Mach-II group that there can be a substantial performance hit, so I'll just do it on a case by case basis. Not sure what the deal is here, since the autowiring only happens one time (at app startup)? What is the "performance hit" being discussed here? It still seems like one should be able to define a base class and inherit from it with properties intact, but there may be something I'm just not "getting" about the Coldspring paradigm. You can define a base class and inherit from it. Public getters and setters for properties in a base class are autowired when you create a subclass (assuming you pass the subclass the appropriate properties in the XML). This seems to be the part that is confusing people. If you set this up correctly it absolutely does work. For now, I think autowiring will work in those cases where I'm using inheritance and I'll explicitly define my mock objects in my unit test xml. Thanks for the link. It validated my confusion and pointed me in the direction of a workaround. I commented again on Doug's blog to explain his error more completely. Hope this helps, Brian
