As a quick note, it does the same kind of high CPU usage on
"WAVELET_SELF_ADDED" events - I removed and re-added it to a wave I'm
working in, and it was 1,944 cpu_ms for that transaction.

On Nov 19, 4:16 pm, "Chris C." <[email protected]> wrote:
> I would show you the code, but I don't know which part to show. I'm
> extracting the blip text and parsing it with a generated parser,
> looking for particular tokens. I am very new at using parsers (a
> different version of this robot uses regular expressions to do
> essentially the same thing) and it may be that using this particular
> one is just too processor-intensive (I didn't write the parser; it's
> generated from a grammar).
>
> I'm not doing any range or annotation stuff with this robot right now,
> though, so that's probably not an issue. However, since the thing that
> changed was using a parser to parse the blip text, as opposed to a set
> of regex's, it's probably the parser.
>
> On the other hand, I am still curious what prompts the
> "beanSerializer" entries to show in the logs sometimes, and other
> times not. Perhaps it's an indicator that for some reason the
> serialization is taking too long.
>
> Thanks for the advice!
>
> -Chris
>
> On Nov 19, 3:54 pm, Olreich <[email protected]> wrote:
>
>
>
> > Look at the Docks, they'll tell you the different functions associated
> > with each of those classes, but pretty much anything you do that
> > modifies waves, blips, annotations, or ranges will have those classes
> > called. It would help to see the code. I personally have never seen
> > that high of a CPU time, but I haven't created a test suite or
> > anything, so I haven't used all of the functions. It could also be in
> > your code, anytime that you are running loops, the potential exists
> > for you to get high CPU, especially if you are playing with
> > Annotations or a Range in that loop (as I believe those are the most
> > intensive operations currently available in the API).
>
> > On Nov 19, 3:11 pm, "Chris C." <[email protected]> wrote:
>
> > > I have a robot I'm experimenting with, and I am frequently - though
> > > not quite always - seeing in the logs report of very high CPU use
> > > during a request, along with the following log messages:
>
> > > com.metaparadigm.jsonrpc.BeanSerializer analyzeBean: analyzing
> > > com.google.wave.api.impl.WaveletData
> > > com.metaparadigm.jsonrpc.BeanSerializer analyzeBean: analyzing
> > > com.google.wave.api.impl.BlipData
> > > com.metaparadigm.jsonrpc.BeanSerializer analyzeBean: analyzing
> > > com.google.wave.api.Annotation
> > > com.metaparadigm.jsonrpc.BeanSerializer analyzeBean: analyzing
> > > com.google.wave.api.Range
>
> > > Those messages are always present when CPU time is high (for instance,
> > > in the neighborhood of 2000 - 3000 cpu_ms), and I've never seen them
> > > in the logs of a request with low CPU use. The robot continues to
> > > work, but obviously I don't want to be slamming the CPU all the time.
>
> > > As mentioned, it's an in-development robot so obviously my code may be
> > > to blame here. Can anyone let me know what prompts those particular
> > > events to happen? Knowing that could help me find mistakes/badthings
> > > in my code, and reduce the CPU load.
>
> > > Thanks,
> > > Chris

--

You received this message because you are subscribed to the Google Groups 
"Google Wave API" 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/google-wave-api?hl=.


Reply via email to