We haven't re-tested recently but the main upshot was that we cached the slowest pages outside of Tapestry by storing the generated HTML on our server. This has meant that we have side-stepped the problem rather than solve it. We never did get to the bottom of the apparent single threading issue.
>From reading around, OGNL seems to be the slowest part of Tapestry. Some of our original components rely very heavily on OGNL both within the HTML and within the actual Java code. It seems that these are our slowest components. Apparently Tapestry 4 doesn't rely so heavily on it so that should improve performance, but I haven't read up on it so that's only what I've heard, not hard fact.
If we do find out any more we'll report back.
Rob Cole
Pedro Abelleira Seco <pe...@usc.es>
26/08/2005 09:26 To
Robert Cole/DMGIT/DMG UK/DeuBa@DMG UK
cc
Subject
Tapestry performance
Hi, Robert.
A couple of weeks ago started a thread in the Tapestry user list about
Tapestry (bad) performance, which seems to be not concluded.
What was the outcome of this?
Thank you very much.
--
------------------------------------------
Pedro Abelleira Seco
Área TIC
Universidade de Santiago de Compostela
15782 Santiago de Compostela (SPAIN)
http://www.usc.es/saus
------------------------------------------
--
This e-mail may contain confidential and/or privileged information. If you are not the intended recipient (or have received this e-mail in error) please notify the sender immediately and destroy this e-mail. Any unauthorized copying, disclosure or distribution of the material in this e-mail is strictly forbidden.
---------------------------------------------------------------------
To unsubscribe, e-mail: tapestry-user-unsubscribe@jakarta.apache.org
For additional commands, e-mail: tapestry-user-help@jakarta.apache.org
holy crap, awesome :) Well, wtf drew? ;) ... On 8/26/05, Mind Bridge <mi...@yahoo.com> wrote: > > Someone with a lot of time on their hands and a clever mind could > > probably speed up all of those ognl calls by turning them into > > javassist or cglib enhancements on the components. > > > I already did that more than an year or two ago and sent some sample > code to Drew. I believe he took the ideas into account when designing > the new version of OGNL. Unfortunately, I haven't had the time to follow > up and help out. > > The speedup achieved back then was around 9 times for OGNL and about 3 > times for Tapestry. I am guessing you would see similar numbers when > using the new OGNL version. > > -mb > > Jesse Kuhnert wrote: > > >Someone with a lot of time on their hands and a clever mind could > >probably speed up all of those ognl calls by turning them into > >javassist or cglib enhancements on the components. Kind of like > >hibernate does. > > > >I don't know anyone who has that kind of time though ;) Maybe > >someone's employer will decide that they absolutely ~must~ have > >tapestry run faster and will fund the development of said > >enhancements. :) > > > >j > >On 8/26/05, Konstantin Ignatyev <kg...@yahoo.com> wrote: > > > > > >>My tests show that Tapestry is definitely slower than > >>Struts+Tiles or Struts+Sitemesh but faster tham > >>JSF(MyFaces). > >> > >>In all cases performance is very adequate. > >> > >> > >> > >>Konstantin Ignatyev > >> > >> > >> > >> > >>PS: If this is a typical day on planet earth, humans will add fifteen million tons of carbon to the atmosphere, destroy 115 square miles of tropical rainforest, create seventy-two miles of desert, eliminate between forty to one hundred species, erode seventy-one million tons of topsoil, add 2,700 tons of CFCs to the stratosphere, and increase their population by 263,000 > >> > >>Bowers, C.A. The Culture of Denial: Why the Environmental Movement Needs a Strategy for Reforming Universities and Public Schools. New York: State University of New York Press, 1997: (4) (5) (p.206) > >> > >>--------------------------------------------------------------------- > >>To unsubscribe, e-mail: tapestry-user-unsubscribe@jakarta.apache.org > >>For additional commands, e-mail: tapestry-user-help@jakarta.apache.org > >> > >> > >> > >> > > > >--------------------------------------------------------------------- > >To unsubscribe, e-mail: tapestry-user-unsubscribe@jakarta.apache.org > >For additional commands, e-mail: tapestry-user-help@jakarta.apache.org > > > > > > > > > > --------------------------------------------------------------------- > To unsubscribe, e-mail: tapestry-user-unsubscribe@jakarta.apache.org > For additional commands, e-mail: tapestry-user-help@jakarta.apache.org > > --------------------------------------------------------------------- To unsubscribe, e-mail: tapestry-user-unsubscribe@jakarta.apache.org For additional commands, e-mail: tapestry-user-help@jakarta.apache.org
> Someone with a lot of time on their hands and a clever mind could > probably speed up all of those ognl calls by turning them into > javassist or cglib enhancements on the components. I already did that more than an year or two ago and sent some sample code to Drew. I believe he took the ideas into account when designing the new version of OGNL. Unfortunately, I haven't had the time to follow up and help out. The speedup achieved back then was around 9 times for OGNL and about 3 times for Tapestry. I am guessing you would see similar numbers when using the new OGNL version. -mb Jesse Kuhnert wrote: >Someone with a lot of time on their hands and a clever mind could >probably speed up all of those ognl calls by turning them into >javassist or cglib enhancements on the components. Kind of like >hibernate does. > >I don't know anyone who has that kind of time though ;) Maybe >someone's employer will decide that they absolutely ~must~ have >tapestry run faster and will fund the development of said >enhancements. :) > >j >On 8/26/05, Konstantin Ignatyev <kg...@yahoo.com> wrote: > > >>My tests show that Tapestry is definitely slower than >>Struts+Tiles or Struts+Sitemesh but faster tham >>JSF(MyFaces). >> >>In all cases performance is very adequate. >> >> >> >>Konstantin Ignatyev >> >> >> >> >>PS: If this is a typical day on planet earth, humans will add fifteen million tons of carbon to the atmosphere, destroy 115 square miles of tropical rainforest, create seventy-two miles of desert, eliminate between forty to one hundred species, erode seventy-one million tons of topsoil, add 2,700 tons of CFCs to the stratosphere, and increase their population by 263,000 >> >>Bowers, C.A. The Culture of Denial: Why the Environmental Movement Needs a Strategy for Reforming Universities and Public Schools. New York: State University of New York Press, 1997: (4) (5) (p.206) >> >>--------------------------------------------------------------------- >>To unsubscribe, e-mail: tapestry-user-unsubscribe@jakarta.apache.org >>For additional commands, e-mail: tapestry-user-help@jakarta.apache.org >> >> >> >> > >--------------------------------------------------------------------- >To unsubscribe, e-mail: tapestry-user-unsubscribe@jakarta.apache.org >For additional commands, e-mail: tapestry-user-help@jakarta.apache.org > > > > --------------------------------------------------------------------- To unsubscribe, e-mail: tapestry-user-unsubscribe@jakarta.apache.org For additional commands, e-mail: tapestry-user-help@jakarta.apache.org
As a general design guideline, I suggest that it's better not to put complex expressions in the html template. Every time I see someone mention an OGNL expression that indexes into an array or does a map lookup, I think it should be moved to a simple getter method in the java code where it's easier to find/debug/log/modify. And a little faster. ----- Original Message ----- From: "Jesse Kuhnert" <jk...@gmail.com> To: "Tapestry users" <ta...@jakarta.apache.org> Sent: Friday, August 26, 2005 10:52 AM Subject: Re: Tapestry performance Someone with a lot of time on their hands and a clever mind could probably speed up all of those ognl calls by turning them into javassist or cglib enhancements on the components. Kind of like hibernate does. I don't know anyone who has that kind of time though ;) Maybe someone's employer will decide that they absolutely ~must~ have tapestry run faster and will fund the development of said enhancements. :) j On 8/26/05, Konstantin Ignatyev <kg...@yahoo.com> wrote: > My tests show that Tapestry is definitely slower than > Struts+Tiles or Struts+Sitemesh but faster tham > JSF(MyFaces). > > In all cases performance is very adequate. > > > > Konstantin Ignatyev > > > > > PS: If this is a typical day on planet earth, humans will add fifteen million tons of carbon to the atmosphere, destroy 115 square miles of tropical rainforest, create seventy-two miles of desert, eliminate between forty to one hundred species, erode seventy-one million tons of topsoil, add 2,700 tons of CFCs to the stratosphere, and increase their population by 263,000 > > Bowers, C.A. The Culture of Denial: Why the Environmental Movement Needs a Strategy for Reforming Universities and Public Schools. New York: State University of New York Press, 1997: (4) (5) (p.206) > > --------------------------------------------------------------------- > To unsubscribe, e-mail: tapestry-user-unsubscribe@jakarta.apache.org > For additional commands, e-mail: tapestry-user-help@jakarta.apache.org > > --------------------------------------------------------------------- To unsubscribe, e-mail: tapestry-user-unsubscribe@jakarta.apache.org For additional commands, e-mail: tapestry-user-help@jakarta.apache.org --------------------------------------------------------------------- To unsubscribe, e-mail: tapestry-user-unsubscribe@jakarta.apache.org For additional commands, e-mail: tapestry-user-help@jakarta.apache.org
Someone with a lot of time on their hands and a clever mind could probably speed up all of those ognl calls by turning them into javassist or cglib enhancements on the components. Kind of like hibernate does. I don't know anyone who has that kind of time though ;) Maybe someone's employer will decide that they absolutely ~must~ have tapestry run faster and will fund the development of said enhancements. :) j On 8/26/05, Konstantin Ignatyev <kg...@yahoo.com> wrote: > My tests show that Tapestry is definitely slower than > Struts+Tiles or Struts+Sitemesh but faster tham > JSF(MyFaces). > > In all cases performance is very adequate. > > > > Konstantin Ignatyev > > > > > PS: If this is a typical day on planet earth, humans will add fifteen million tons of carbon to the atmosphere, destroy 115 square miles of tropical rainforest, create seventy-two miles of desert, eliminate between forty to one hundred species, erode seventy-one million tons of topsoil, add 2,700 tons of CFCs to the stratosphere, and increase their population by 263,000 > > Bowers, C.A. The Culture of Denial: Why the Environmental Movement Needs a Strategy for Reforming Universities and Public Schools. New York: State University of New York Press, 1997: (4) (5) (p.206) > > --------------------------------------------------------------------- > To unsubscribe, e-mail: tapestry-user-unsubscribe@jakarta.apache.org > For additional commands, e-mail: tapestry-user-help@jakarta.apache.org > > --------------------------------------------------------------------- To unsubscribe, e-mail: tapestry-user-unsubscribe@jakarta.apache.org For additional commands, e-mail: tapestry-user-help@jakarta.apache.org
My tests show that Tapestry is definitely slower than Struts+Tiles or Struts+Sitemesh but faster tham JSF(MyFaces). In all cases performance is very adequate. Konstantin Ignatyev PS: If this is a typical day on planet earth, humans will add fifteen million tons of carbon to the atmosphere, destroy 115 square miles of tropical rainforest, create seventy-two miles of desert, eliminate between forty to one hundred species, erode seventy-one million tons of topsoil, add 2,700 tons of CFCs to the stratosphere, and increase their population by 263,000 Bowers, C.A. The Culture of Denial: Why the Environmental Movement Needs a Strategy for Reforming Universities and Public Schools. New York: State University of New York Press, 1997: (4) (5) (p.206) --------------------------------------------------------------------- To unsubscribe, e-mail: tapestry-user-unsubscribe@jakarta.apache.org For additional commands, e-mail: tapestry-user-help@jakarta.apache.org