cocoon-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From hussayn <>
Subject Re: Preparing the new dojo-rsrc build
Date Fri, 12 Sep 2008 07:09:35 GMT

Grzegorz Kossakowski-2 wrote:
> hussayn pisze:
>> Anyways, i realised at some time of my experiments, that we got some
>> performance issues in our production environment. So we eventually
>> abandonned the above mentioned dojo-block and placed 
>> the plain vanilla dojo-distribution directly on our apache-frontend
> Hussayn, could you elaborate on problems you had?
>> so that
>> the dojo-files are now directly served from the web-server as static
>> files
>> (and we can simply upgrade to newer dojo-versions without modifying a
>> cocoon-block...)
> Dojo should be packaged into separate jar, so upgrading Dojo should be as
> easy as replacement of 
> this single jar which looks like not much more difficult than replacing
> directory for httpd, right?

Hi, Grzegorz

I got a weird effect, which i still can not fully understand, but its
something like this:

when i developed (on windows-xp, jetty running on localhost) everything was
fine for me, reaction time was quite satisfying.

Then i switched to our testserver (suse-10.0 plus apache2-frontend). ok, the
test server stands at a remote location, but it has 2 2Mbit connection, so i
do not expect any performance issues due to internet connectivity problems.
But unexpectedly now my application took more than 15 seconds to show up on
the screen (everytime i call the page).
That was when i detected, that i get about 30 files from dojo, needing about
half a second per file.
On top of this, dojo now seems to be much slower, the browser repeatedly
freezes for about 10 seconds, before the dojo-screen shows up.

So i downloaded the dojo-distribution to our apache2-server and reconfigured
my application so that it calls the dojo-files directly from our webserver.
By now my application reacted 3 times faster, buit still by far slower
compared to the localhost installation, still i see dojho-freezes, but the
resource files get in much faster.

So to conclude, i guess this is a cacheing problem ? I did not expect to see
the dojo-resource files travel over the line again and again. Shouldn't the
browser already take the resources out of its own cache ?
Maybe... while the packages travel from the server to the browser at some
place, cacheing information gets discarded ? and thats why my local
application seems to behave soo much faster ? Could that be ?
And how could i debug that ?

regards, husssayn 
View this message in context:
Sent from the Cocoon - Dev mailing list archive at

View raw message