[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
Re: [aspectj-users] compile time performance (Eclipse 3.3)

Hi Andy,

as Andrew wrote Eclipse 3.4 would be the answer of our problem (and I had expected this answer;-). We have it still on our todo list but the last time we tried it we failed. The problem seems to be the Webtools - our JSF application fails to start with strange classpath problems. I guess we have to setup the Eclipse project from scratch but we were not sure if this solve the problem. So we decided not to do it in the past.

A backport to Eclipse 3.3 would be nice but since we really want to use Eclipse 3.4 this would probably not necessary for us. I think now it is a good time to try again an update to Eclipse 3.4 / WTP 3.0.


Andy Clement schrieb:
Hi Oliver,

The last few times I've been involved in determining why a build is slow, it has turned out to be nothing to do with the aspects or the weaving. So far it has always proved to be incremental compilation: the analysis of what needs to be built 'this time', and the downstream effects of that change (what other types need rebuilding because of what we just did, and which other projects need a rebuild).

So, to progress this - I need to know more details about it - is it the first full build of your project that depends on the aspectj lib that is slow? is it the incremental builds of the project using your aspectj library? Have you tried opening the AJDT Event Trace view and see where the time goes during a build? It is best to raise a bug to report these things or I'll lose the details in my inbox, but the end result might be that I tell you that you need to upgrade to Eclipse 3.4 to address it. We have made huge changes in incremental analysis during the recent releases of AJ and AJDT. We have an open bug to backport some of the best benefits back to Eclipse 3.3 (and create an AJDT 1.5.4) but I haven't seen any votes or feedback on that bug indicating people want it. It would not be a trivial piece of work so we need real justification to put the effort in.

AJDT 1.6.4 may not be the complete answer, but it ought to be a step in the right direction. We are currently progressing incremental build issues under bug https://bugs.eclipse.org/bugs/show_bug.cgi?id=268827 - all the changes for which will hopefully be in AspectJ 1.6.4 (then AJDT 1.6.5).


2009/3/21 Oliver Böhm <boehm@xxxxxxxxxx <mailto:boehm@xxxxxxxxxx>>


    we use Eclipse 3.3 / ADJT 1.5.3 with Webtools in a project. We had
    big problem with one AspectJ lib (patterntesting-check). It has some
    general aspects which seems to slow down the compile time a lot. My

    * Is the compile time performance of Eclipse 3.4 / AJDT 1.6.3 much
    better than with AJDT 1.5.3?

    * Are there any "best practices" or performance tips available how
    to use AspectJ?

    * Is it possible to profile the AspectJ compiler to see, which
    aspects slow down the compile time?


    aspectj-users mailing list
    aspectj-users@xxxxxxxxxxx <mailto:aspectj-users@xxxxxxxxxxx>


aspectj-users mailing list

-- Oliver Böhm http://www.javatux.de