| Re: [aspectj-users] Runtime performance against different weaving options |
|
Cool, Thanks a lot!
Leon > From: alexander@xxxxxxxxxxxxxx > To: aspectj-users@xxxxxxxxxxx > Date: Wed, 4 Jun 2014 11:18:48 +0200 > Subject: Re: [aspectj-users] Runtime performance against different weaving options > > Hi Leon. > > In general, CTW and PCTW should be the same in performance because you just have normal classloading plus aspectjrt.jar. Only for LTW you have a "warm-up" phase because aspect weaving needs to be done during classloading. This is why you need the weaving agent (aspectjweaver.jar which also includes the runtime) on the command line. So LTW is a bit slower at the beginning, afterwards the three methods should have identical performance. Either way runtime performance will be significantly faster (not mentioning more powerful) than with proxy-based AOP approaches like Spring AOP which involves Java Dynamic Proxies and/or CGLIB proxies. > > Regards > -- > Alexander Kriegisch > http://scrum-master.de > > > 马leon schrieb am 04.06.2014 11:10: > > > I know there're 3 ways to weaving: compile-time, post compile time and load-time > > > > I'd like to know is there any performance comparison for above 3 ways. > > > > By "performance", I mean all classes have been loaded and the server gets warm-up for some time. > > _______________________________________________ > aspectj-users mailing list > aspectj-users@xxxxxxxxxxx > https://dev.eclipse.org/mailman/listinfo/aspectj-users |