Most of today was spent doing nothing much interesting. I watched a PC that I was working just randomly reboot (yay for Windows, what a wonderful operating system), I did manage to get some extra documentation written on my SSO plugins but most interestingly we’ve now got our ADSL connection back so we’re not trying to share a dialup internet connection with three people. It was very painful. Its nice to have pings that aren’t in the 10 to 14 second range (I kid you not). I’ve spent most of this evening writing an OJS plugin for my girlfriend  – she keeps changing the goal posts each time she asks me which has now lead to a minor redesign of the way I’m approaching things, I’m going to have to spend a lot more time digging through OJS to find the code that I need to do to get the system to behave properly but sooner or later she wants me to hack apart the system to change it anyway.

Today was marked by a rather tedious project stocktake and validating that all projects are in version control. I picked up two services that weren’t properly versioned, but all of the other projects appear to have been fine though some had a few changes in my local workspace that needed to be checked in but all in all most projects were up to date already. I did find some projects that need some attention, some that need to be updated from 1.0 to 1.5 and some that are ready to be released properly that are sitting locked up not doing much. So next week will hopefully result in a few new releases.

Today was marked by my completion of my token login project (now working properly!) and playing with the internet creating dial up NAT’d networks and playing with IPv6 and 6to4 services.

Read more

Some days I really worry about Telstra, Australia’s largest telecommunications provider. We’ve just moved into a new unit and when it was confirmed we had the place from the agent my girlfriend called Telstra to schedule getting the phone put on. Over the weekend we moved and ended up being faster than expected so on monday she rang Telstra to bring forward the date and was told after 20 minutes that the unit didn’t even have a phone line and that he couldn’t find the previous order either. He said that it’s cost some large amount of money to get a phone line. At this point we both headed off to work after quickly checking with the agent that they thought there was a phone line (they thought there should be). Later on we unpacked the phone and connected it up to be greeted by a dial tone. This morning we rang Telstra who then found both orders in the “old” system and we were told that it would be connected in an hour. Now it’s time to get a net connection and time will tell with that.

On the work front I spent some time cleaning up the cart functionality and training the users on how to use the mail merge functionality with the system. They’ve gotten someone in who isn’t retired to help out on the computing side which will hopefully improve their capability to work without needing to contact us. With any luck we won’t have too many more issues.

Lastly I spent a small amount of time to build a token login system. At the moment it has a simple library to issue, revoke and validate tokens as well as a SSO plugin to detect logins. It defaults to alowing 5 logins with the token and is valid for five days before it expires. Before I go much further with it I’ll need to build a component to revoke tokens and create the database table. This shouldn’t be too hard to do and will expand on the entire JAuthTools collection.

In part I understand part of agile development requires spend some, if not all of your time, with the client. The majority of my time was spent on site with the users working through issues with our system architect. Today’s major issue is printing replacement labels for lost or missing ballots. So far they have been manually retyping in information. We have managed to progress to getting the labels to the point that they are being copied, but hopefully by the end of the week we’ll have a much more scalable solution as we start getting busier. Can’t wait until this is over and I can go back to normality.

This week marks the start of my election systems into production and inevitably some scope creep. As the system architect commented that the project never really had any scope. The actual users of the system have different expectations and requirements than we have encountered so far. The main difference is the request for a very simple direct search system instead of the very flexible system that I had put into the system. A quick modification locally resolved the issue for them and when I returned to my desk I altered the system to handle both the search requirements that they were interested in whilst also providing the more advanced search as an option. This and other smaller changes took up the majority of my day in addition to a bit of training and on site monitoring.

Friday is always a slow day for most and today was no exception. That said I managed to be productive.

I had a quick play with a PHP task tracker called TaskFreak which seems to be a nifty web based tracker and is open source. Hopefully it will turn out to be a useful tool in the long run. It appears really smooth with a nice Web 2.0 style interface but doesn’t appear to be AJAX driven from the small amount of playing but it might be used tastefully. So maybe more about it later.

I also spent some time working on the Defect Register adding the ability for it to handle documents for a defect. The defect register is a perfect example of how not to do a PHP web app – it’s the sort of poorly written and designed application that the Python people pick up and declare the language to be completely useful and insecure. But that aside, it’s painful to alter at times and it took me a bit longer to alter than I would like whilst reminding me just how bad and ugly the code really is in places. As I do everytime I get a chance, I managed to rewrite a little bit and add tools that I can use later. The users appear to be happy with the new functionality so we’ll see.

Lastly I did some more work on my new SSO framework and even did some initial documentation work on my wiki. I’ve decided that for the weekend I’ll build the token login system that I’ve been meaning to do as part of my university work. Hopefully I’ll get a part of my SSO framework built whilst doing this as well.

I spent the majority of today working on non-technology related projects (lots of face time with people) though I did spend some time looking how I handle SSO and different SSO options in Joomla!. Now my definition of SSO usually differentiates from most people because I offer two different authentication options. To me I define “Single Sign On” (SSO) as when you sign on once, and that then gets used everywhere else. A great example of this is Microsoft’s Integrated Windows Authentication where you sign onto your computer once and then everything authenticates you from there without having you to enter your username or password for the entire session (Microsoft achieves this by using “Kerberos” which is available for all platforms in various solutions). This is then juxtaposed with “Single Sign In” where you use the same username and password but type it into each service individually. The reason for creating this distinction was for when I was working in Joomla! 1.0 on two plugins: one that would allow you to just login using details stored in LDAP and another that would automatically log you in from LDAP – historically both identified as “SSO”, however two plugins called “LDAP SSO” gets a tad confusing especially since the LDAP SSO solution was in fact tied strongly to eDirectory, so one became SSI since you had to redivulge your credentials to it and the other became SSO since it automatically worked you out using other details.

In Joomla! 1.5, my LDAP SSI plugin became the “Authentication – LDAP” plugin and I wrote an updated version of the SSO system now splitting it into two parts: one for identifying the user (the SSO part which returns a username) and one for user sources (e.g. finding details about a given username). So far the SSO system is bootstrapped off a system plugin and the user source system is subservient to it. This allows a disconnect between the parts of the system that work out who the person is and finding their details. Of course, not all systems handle this well – some SSO systems also find out details useful for the user source system (e.g. where information is stored in the SSO packet) so there is also a session based user source plugin that allows the SSO system to provide details without requiring an extra user source plugin. In the next release of JAuthTools there will be a new authentication plugin that features the ability to do everything the built in plugin does as well as integrating with the user source system and providing the ability to require that a user existing in Joomla!. Another new feature is a SSO bootstrap module instead of the system plugin but today is something that I’m trying to work out how to do well: build a generic SSO component.

Now at one point the organisation that pays me money (yay!) needed the ability to transition users automatically between two Joomla! sites. I found an older system that did this for Joomla! 1.0 (though it didn’t work properly, it was only a minor modification to get it to work) and at the time the requirement was actually between a Joomla! 1.0 and 1.5 site so I fixed it up to work with both Joomla! 1.0 and 1.5 legacy mode. Now I haven’t done much with that since I built it, but I’d like to take it and update it to something that is 1.5 native and integrates a lot better. So I’m looking at rethinking the way that it works at the moment and redesigning some of the other systems that I have floating around (such as the SimpleSSO system that I wrote to handle Google authentication better). The thought I’m having is making two categories of SSO system (perhaps finding a name for each): one of the more traditional ones based on Kerberos and others based on OpenID, the SOAP SSO tool I’ve co-opted along the way and my own SimpleSSO system. The difference is that one system can authenticate the user easily in one request or a second request (issuing an authentication challenge, potentially handled at the server level above PHP) and the other typically involves the user bouncing around the web testing if the user is authenticated, if the user trusts the site and then authenticating the user to the site probably with another request sent by the server to the remote site. So I’m working out how to integrate those systems in an effective manner. Some of them are just a matter of generating links to the authentication source and others require some nominal user input (e.g. OpenID). I think I’m going to hijack the plugin infrastructure’s XML files to handle stuff and work from there. Now time to build the entire system!

As noted yesterday I have finished up the voter application so today I did the finalisation work to progress it to the next level. So today I spent some time working on removing functionality for the short term deployment where it only needs to be able to search for voters and their address and not their voting status, batch or voting statistics.

As I’ve been blogging lately, I spent another day working on the election system. Today I finalised the system with it being completed a day ahead of plan which is a good feeling. Tomorrow will be a run through with one of the guys looking after the election to make sure that it does what they need. It now has voter search and batch listing (batches are the quanta they use to count votes) with both batch and voter detail views. Finally, there is a statistics view for the number of returned votes and the break down of votes per state that they can be in.