[Catalyst] Catalyst test server response times.

Ian Docherty catalyst at iandocherty.com
Wed Jan 30 16:41:48 GMT 2008


Jason Kohles wrote:
>
> On Jan 30, 2008, at 10:23 AM, Roderick A. Anderson wrote:
>
>> Ian Docherty wrote:
>>> I have moved (back) to using the built in Catalyst server for 
>>> testing purposes.
>>> It is frequently (but not always) taking 20 seconds or so to display 
>>> a page in my browser
>>> even though the timings reported by the Catalyst server are about 
>>> 0.2 sec. so I know that
>>> it is not the Catalyst app that is causing the delay.
>>
>> Having read the other postings on this I wonder if you have a 
>> large-ish css or Javascript file being loaded.  I've noticed a bit of 
>> a delay when I was trying the Ext2.js examples from the Ext2 site.
>
> This delay is related more to the complexity of the javascript than 
> it's size.  It's essentially the compile-time penalty of the library 
> in use, ExtJS is one of the bigger ones that I've noticed it on, it 
> can take quite a while for it get loaded/decompressed/setup and ready 
> to go.  It's much faster when using a release version rather than a 
> testing version, and there are tools that will let you build custom 
> ExtJS versions with just the features you are actually using included, 
> so you don't have to wait for initialization for features you aren't 
> using.
I am using extjs and realised that this could be an issue but I created 
a brand new minimal Catalyst application and I saw the same issues.

Also I used wget to get the web page, which of course does not execute 
the javascript.

The 15 second long delay is prior to the Catalyst debug information 
being printed. The actual execution of the various controllers is very 
quick (sub 0.3 seconds)

If I use CATALYST_DEBUG=0 then the problem goes away.

Lately, the problem has gone away even without turning debug off.

One possibility I am still investigating is that the problem only occurs 
if two or more developers are running a server at the same time. (jury 
is still out on this)

Regards
Ian




More information about the Catalyst mailing list