You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@cocoon.apache.org by Carsten Ziegeler <cz...@sundn.de> on 2001/04/10 17:07:05 UTC
[C2]: Status report on caching
Hi,
at last I had time to start the implementation...
Rather than telling you my proposal I will try to finish the
work tomorrow and show you the result.
The first step is done, caching of the generator phase works.
Now I have to clean up the code and make further improvements.
Carsten
---------------------------------------------------------------------
To unsubscribe, e-mail: cocoon-dev-unsubscribe@xml.apache.org
For additional commands, email: cocoon-dev-help@xml.apache.org
Re: AW: [C2]: Status report on caching - update
Posted by Conny Krappatsch <ck...@virbus.de>.
On Wednesday 11 April 2001 17:02, Carsten Ziegeler wrote:
> > Von: Conny Krappatsch [mailto:conny@smb-tec.com]
> >
> > Quoting Carsten Ziegeler <cz...@sundn.de>:
> > > The storing of the sax events is implemented by two more Avalon
> > > components
> > > the XMLSerializer and the XMLDeserializer. I just used Stefano's
> > > XMLCompiler
> > > and modified it a little bit.
> >
> > On Wednesday 11 April 2001 15:00, Giacomo Pati wrote:
> > > Don't worry but :) I've had an error when I was requesting the
> >
> > jakarta site
> >
> > > in the samples (no data).
> >
> > Maybe this is related to a bug in Stefanos original code, which caused a
> > ArrayIndexOutOfBoundsException to be thrown, if the XML document
> > contains a
> > none ASCII character.
> >
> > I don't know if it has been fix since i reported it to the list:
> > <http://marc.theaimsgroup.com/?l=xml-cocoon-dev&m=97317213022512&w=1>
>
> OK, I will look into it then.
I had a look at org.apache.cocoon.xml.CompiledXMLInputStream.java
and - hmmm - it's fixed - somehow. The (unexpected ) behaviour is now
described in the javadoc comment of readChars().
I was wrong with the ArrayIndexOutOfBoundsException. The problem was (is?),
if the XML document contains none ASCII characters, getChars() returns a
character array which has 0x00 characters at the end. This may lead to
strange errors if you don't expect it.
Conny
---------------------------------------------------------------------
To unsubscribe, e-mail: cocoon-dev-unsubscribe@xml.apache.org
For additional commands, email: cocoon-dev-help@xml.apache.org
AW: [C2]: Status report on caching - update
Posted by Carsten Ziegeler <cz...@sundn.de>.
> Von: Conny Krappatsch [mailto:conny@smb-tec.com]
> Quoting Carsten Ziegeler <cz...@sundn.de>:
> > The storing of the sax events is implemented by two more Avalon
> > components
> > the XMLSerializer and the XMLDeserializer. I just used Stefano's
> > XMLCompiler
> > and modified it a little bit.
>
> On Wednesday 11 April 2001 15:00, Giacomo Pati wrote:
> > Don't worry but :) I've had an error when I was requesting the
> jakarta site
> > in the samples (no data).
>
> Maybe this is related to a bug in Stefanos original code, which caused a
> ArrayIndexOutOfBoundsException to be thrown, if the XML document
> contains a
> none ASCII character.
>
> I don't know if it has been fix since i reported it to the list:
> <http://marc.theaimsgroup.com/?l=xml-cocoon-dev&m=97317213022512&w=1>
>
OK, I will look into it then.
Carsten
---------------------------------------------------------------------
To unsubscribe, e-mail: cocoon-dev-unsubscribe@xml.apache.org
For additional commands, email: cocoon-dev-help@xml.apache.org
Re: [C2]: Status report on caching - update
Posted by Conny Krappatsch <co...@smb-tec.com>.
Quoting Carsten Ziegeler <cz...@sundn.de>:
> The storing of the sax events is implemented by two more Avalon
> components
> the XMLSerializer and the XMLDeserializer. I just used Stefano's
> XMLCompiler
> and modified it a little bit.
On Wednesday 11 April 2001 15:00, Giacomo Pati wrote:
> Don't worry but :) I've had an error when I was requesting the jakarta site
> in the samples (no data).
Maybe this is related to a bug in Stefanos original code, which caused a
ArrayIndexOutOfBoundsException to be thrown, if the XML document contains a
none ASCII character.
I don't know if it has been fix since i reported it to the list:
<http://marc.theaimsgroup.com/?l=xml-cocoon-dev&m=97317213022512&w=1>
regards,
Conny
---------------------------------------------------------------------
To unsubscribe, e-mail: cocoon-dev-unsubscribe@xml.apache.org
For additional commands, email: cocoon-dev-help@xml.apache.org
Re: [C2]: Status report on caching - update
Posted by Giacomo Pati <gi...@apache.org>.
Quoting Carsten Ziegeler <cz...@sundn.de>:
> Hi Cocooners,
>
> let's do a "Calling all stations"...
>
> ...as the first caching implementation is just finished!
> Try it out and simply have fun.
>
> For the caching to work you have to set up the CachingEventPipeline
> in the cocoon.xconf as the EventPipeline.
>
> The first tests for the welcome, the html, voxml and pdf pages
> where delivered from the cache between 4-6 times faster than fresh
> generated.
THIS IS TOTALLY AMAZING :)
I've dropped everything I was doing only to test it!
It's really cool. The overhead the SAX "tee"ing has is rawly about 10-20% on the
first request (well, I've not done very carefull test so far). Afterward it
speeds up easily about 5 times.
Don't worry but :) I've had an error when I was requesting the jakarta site in
the samples (no data).
It's a great start for it. Well done, Carsten.
> Concept
> -------
> The concept of the caching is based on the various discussions and
> my own thoughts...So here only a short overview:
>
> The sitemap components (currently Generator and Transformer) can
> implement the Cacheable interface. This interface has two method:
> - generateKey() to generate a unique key within the space of the
> component ,e.g for the FileGenerator this key is a hash of the
> source argument.
> (For this I added the HashUtil class which provides the BuzHash
> algorithm).
> - generateValidity() the validity which checks if the cached content
> is still valid, e.g. for the FileGenerator this is the TimeStamp
> of the xml file.
>
> The Caching Algorithm now checks all pipeline components for
> the Cacheable interface and builds a unique key from these Components
> (starting from the generator just to the point before the first not
> Cacheable component).
> This key is used to get a CachedObject from the EventCache (actually
> a Store). The CachedObject (if available) contains the cached content
> and all validity objects when this was cached.
> The next stage compares the current validity objects with the cached
> ones. If they are all valid the cached response is used.
> If something failes during this algorithm, a new response is generated
> and this result is cached.
>
> The storing of the sax events is implemented by two more Avalon
> components
> the XMLSerializer and the XMLDeserializer. I just used Stefano's
> XMLCompiler
> and modified it a little bit.
>
> "That's all".
>
>
> What is left to do (regarding caching)?
> ---------------------------------------
> - Further cleanup of the code (I will do that in the next days)
> - Make more components Cacheable
> - Proove if this concept is sufficient.
> - Check if we still need the cache efficiency algorithm
> - Integrate ContentAggregation
> Please explain what me what you need for caching ContentAggregation.
My latest vision about content aggregation was to have a semantic in the sitemap
to express which cocoon internal request (represented as EventPipelines) makes
up the initial content of the page. Those collected (sub) EventPipelines would
be connected in turn to the main EventPipeline. If those sub EventPipeline can
act like Generators and express their cachability it could eventually be
possible to cached it as a hole.
> I totally ignored this part for the first implementation. I don't use
> any
> SAXConnectors stuff - as I do not understand the sense of them right
> now.
It's meant as transparent xinclude transformers.
> But I am sure that together, we can solve this.
I'm very confident :)
> We should solve this all in the next week.
I've planned to work tomorow on the CA stuff.
> Wish you all a happy easter!
Same to you.
Giacomo
---------------------------------------------------------------------
To unsubscribe, e-mail: cocoon-dev-unsubscribe@xml.apache.org
For additional commands, email: cocoon-dev-help@xml.apache.org
[C2]: Status report on caching - update
Posted by Carsten Ziegeler <cz...@sundn.de>.
Hi Cocooners,
let's do a "Calling all stations"...
...as the first caching implementation is just finished!
Try it out and simply have fun.
For the caching to work you have to set up the CachingEventPipeline
in the cocoon.xconf as the EventPipeline.
The first tests for the welcome, the html, voxml and pdf pages
where delivered from the cache between 4-6 times faster than fresh
generated.
Concept
-------
The concept of the caching is based on the various discussions and
my own thoughts...So here only a short overview:
The sitemap components (currently Generator and Transformer) can
implement the Cacheable interface. This interface has two method:
- generateKey() to generate a unique key within the space of the
component ,e.g for the FileGenerator this key is a hash of the
source argument.
(For this I added the HashUtil class which provides the BuzHash
algorithm).
- generateValidity() the validity which checks if the cached content
is still valid, e.g. for the FileGenerator this is the TimeStamp
of the xml file.
The Caching Algorithm now checks all pipeline components for
the Cacheable interface and builds a unique key from these Components
(starting from the generator just to the point before the first not
Cacheable component).
This key is used to get a CachedObject from the EventCache (actually
a Store). The CachedObject (if available) contains the cached content
and all validity objects when this was cached.
The next stage compares the current validity objects with the cached
ones. If they are all valid the cached response is used.
If something failes during this algorithm, a new response is generated
and this result is cached.
The storing of the sax events is implemented by two more Avalon components
the XMLSerializer and the XMLDeserializer. I just used Stefano's XMLCompiler
and modified it a little bit.
"That's all".
What is left to do (regarding caching)?
---------------------------------------
- Further cleanup of the code (I will do that in the next days)
- Make more components Cacheable
- Proove if this concept is sufficient.
- Check if we still need the cache efficiency algorithm
- Integrate ContentAggregation
Please explain what me what you need for caching ContentAggregation. I
totally ignored this part for the first implementation. I don't use any
SAXConnectors stuff - as I do not understand the sense of them right now.
But I am sure that together, we can solve this.
We should solve this all in the next week.
Wish you all a happy easter!
Carsten
Open Source Group sunShine - b:Integrated
================================================================
Carsten Ziegeler, S&N AG, Klingenderstrasse 5, D-33100 Paderborn
www.sundn.de mailto: cziegeler@sundn.de
================================================================
---------------------------------------------------------------------
To unsubscribe, e-mail: cocoon-dev-unsubscribe@xml.apache.org
For additional commands, email: cocoon-dev-help@xml.apache.org