Lilina News Aggregator

Google has officially stopped allowing new signups for their SOAP search API which was used in Lilina 0.7. Although we have removed it from 1.0, we still get many people asking why they are getting errors about the soapclient class and nusoap.php

The reason behind this error is that in PHP 5, a built in SOAP client was added with the same class name. This has effected other PHP applications which used the Nusoap class.

The simple fix for this is to comment out the line that says require_once './inc/google.php' ;
Simply add // in front of this line and it should work beautifully.

I’ve gone through and removed all accounts on the blog here. They aren’t needed for commenting and as there’s no other reason for them, I’ve disabled signups as well.

I will be setting up integration with the forums so that if you are logged in there, your name, email and URL will be autofilled in the comments form. This should help with a few things I’ll be setting up soon, such as a central place to subscribe to feeds so sites can add “Subscribe to Lilina” buttons.

I have just uploaded Lilina 1.0 Alpha 1 to Google Code. This is taken from revision 100 from Subversion.

Download it now.

I added jQuery into the mix today. It should be able to replace all of the functions currently used in engine.js

As an example, expanding and collapsing items. Up until the revision 89, around 45 lines were used for working out what was clicked and changing it and the background. It took every click and checked whether it was needed. If it was, it went through complicated processes working out what to do.
With jQuery, it’s all handled for us, and the only code we need is:
$(document).ready(function() {
$(".title").click(function(){
$(this).next().next().slideToggle().parent().toggleClass("c1").toggleClass("c2");
});
});

Review: We cut down 45 lines to 3 lines! Everything is handled for us now by jQuery and is about 4 times faster. Plus, it’s ultra easy to add new features, like the removal of date headers. Thanks jQuery!

I now have the cache working. Seriously this time. It now also adds a comment to the top of the page with either Retrieved from cache or Generated fresh. This should aid in debugging.

I now have the “show only from x hours ago” buttons working again.

For anyone following the development of Lilina on SVN, I now have the feed’s source displaying properly on each item.
The installer is also almost finished.

I just tested Lilina on my laptop which had a few programs running, just to get a fairly accurate time on the loading (my computer is bogged down).

The results were amazing:

• 3.12 seconds on the first go
• 1.07 seconds on reload

These times are incredible compared to old loading times of 10 seconds for one feed.

There are some important things to note however:

• HTMLPurifier uses a lot of CPU and memory resources
  • As the author says: “Precisely why you need to cache HTMLPurifier’s output”
  • However, Lilina does not cache HTMLPurifier’s output. This must be fixed
• Lilina isn’t finished yet, so the times may be inaccurate.
• Loading times are dependant on the speed of the server’s connection:
  The faster the server, the less time needed to load

Now, once I get the OPML import up and running, I will be able to test for security flaws.

I finally have the caching system working. I accidentally had it before the settings, which screwed with it because the times weren’t set.

Kellan has released his rewritten code of Magpie.