beehive-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Xibin Zeng" <xibin.z...@gmail.com>
Subject Re: Perf fix in caching of ControlAnnotationProcessor
Date Wed, 05 Sep 2007 16:11:59 GMT
Jacob -

Long time no see! How is everything? Can't believe that I have been Seattle
for over a year now. My new gig is in Belltown on 4th street.

Take care,
Xibin

On 9/4/07, Jacob Danner <jacob.danner@gmail.com> wrote:
>
> Hi All,
> It was suggested to me by a colleague that the performance of apt and
> the ControlAnnotationProcessor could be increased if caching was
> enabled differently. I began to investigating caching of the
> VelocityGenerator and later the ControlAnnotationProcessor. After
> tinkering around, I found that performance could be increased with a
> small change to the ControlAnnotationProcessorFactory to cache the
> instance of the ControlAnnotationProcessor.
>
> To test I was using Workshop for Weblogic with a webproject using the
> beehive-controls facet.
> some not-so-precise perf metrics show the following:
> No Changes
> First Build - ~30 s
> Subsequent Builds - ~25s
>
> Caching of ControlAnnotationProcessor (item4)
> First Build - ~28s
> Subsequent Build - ~23s
>
>
> The change is fairly simple (here it is below):
> public AnnotationProcessor getProcessorFor(Set<AnnotationTypeDeclaration>
> atds,
>
> AnnotationProcessorEnvironment env)
>     {
>         if(_twoPhaseAnnotationProcessor == null)
>             _twoPhaseAnnotationProcessor = new
> ControlAnnotationProcessor(atds, env);
>
>         return _twoPhaseAnnotationProcessor;
>     }
>
>     TwoPhaseAnnotationProcessor _twoPhaseAnnotationProcessor;
>
>
>
> and I've attached a small patch in case anyone is interested in checking
> it out.
> Thanks,
> -Jacob Danner
>
>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message