Jeroens blog

I just released Maps and Semantic Maps 0.2. Both the extensions documentation has been updated. I don’t have time to describe any of the improvements made since version 0.1, but the change logs are linked below.

• Maps change log
• Maps documentation
• Maps demo’s
• Maps downloads

• Semantic Maps change log
• Semantic Maps documentation
• Semantic Maps demo’s
• Semantic Maps downloads

Today I finished refactoring the parser function handling classes (in Maps) for each service and also fixed the php/js logic issue. After that I took care of a variety of to-do’s, including the centre parameter, support for all OpenLayers controls, the display of title’s and labels for markers and the adding of the Google Maps ‘physical’ map type to the map type selector control when it’s set. Here you have a nice screenshot demonstrating the last 2 features:

{{
#display_address:
service=googlemaps
|address=Moscow, Russia
|width=800
|Height=500
|type=physical
|title=Moscow (Russia)
|label=Moscow is the capital and the largest city of Russia. It is also the largest metropolitan area in Europe.
}}

There are only 2 to-do’s remaining before I can release version 0.2 of both extensions: multi-geocoder support and OSM base layers for OpenLayers maps. Both are proving to be quite a challenge though. I was planning to create a central system for the geocoders, both the ones used in parser functions, and in form inputs, but this is proving to be extremely difficult to achieve. I therefore will probably just go for separate systems (unless I find an easy way to do the original plans).

Yay – I almost thought I could finally access my bank account for the first time in what must now almost be a year – of course I couldn’t.

I’ll start at the beginning, going through all the ways their system failed.

All my problems started off with me just trying to sign in to my on-line banking account. Back then they where switching between a system with just logins, and one with a card-reader you need to put your bank card into. Like I always had done, I tried to sign in to my account, but entered a wrong password 3 times according to the site (very strange since I have 100′s of logins, and practically never enter a wrong password – and my portable password manager certainly doesn’t). So I had to wait a little over a week until they send me a new code. When entering it, it failed again (and so the cycle repeated itself a few times).

At some point the people at the banking office got so annoyed (since this was all my fault), they gave me a card-reader so I could use the new system. I have logged in to that once just to checked how it worked. A week later, when I needed to do some transaction, I did exactly the same, but got some obscure error when I put my card into the reader. After sending an email to my bank and waiting a few days, I got the reply the chip was damaged, and I should go get a new card. I went passed the office again, and requested a new card, which I immediately got. I had to wait for a code to be sent though. Due to some delays at the main office, this took approximately 3 weeks (omg?).

Well, when I finally got that code, I of course tried it out right away. Apparently I had to ‘activate’ my card first at an ATM though. I don’t know how, but I somehow failed to sense in the force I had to do this, and only found out after my card got blocked and someone informed me about it. Great, now I had to wait for another code to be sent!

This one only took a week or so, and I managed to activate it at the ATM. (Why is that required – lame analogue nonsense!) So today I tried to login again. I went to the correct section of the Argenta site, entered my details, put in a challenge and my pin into the card reader, and typed the response in the form on the web-page. I didn’t get any errors saying my code was wrong or that the chip of my card was damaged, so I thought ‘woooot, finally!’.

Half a second later, I got this popup:

This is Dutch, translated into English it says: There is a problem with your card. You can contact the support ….

What?! There is A PROBLEM?!!!! with my card????

I send another mail requesting the reason for this. If they fail to give decent support, I’m going to transfer all the money I got there to one of my other accounts at another bank. I’m not going to quit to try get access to my account. It’s just funny how they keep fucking their system up. If this wasn’t Dutch – I’d submit it to failblog. I’m also staring to think it’s just easier to go hack their servers then to access my account in a regular fashion, it would s not surprise me if the security fails as much as their login system.

Since the 0.1 release of Maps and Semantic Maps, I’ve gotten quite some feedback on what features are in demand and what could be improved. Although I’ve not gotten any specific feedback on the extensions source code itself, I decided that refactoring it to get it as high quality as possible was probably the most important thing in the 0.2 release, since this would facilitate adding functionality and making changes to the existing features. The last few days I had some great fun with the refactoring.

I started off with the form inputs in Semantic Maps. I changed the SMFormInput class, which gets inherited by the mapping service specific classes, by refactoring a whole bunch of common code up from the specific classes. I also went for a more OOP approach, and at some point got stuck cause I could not make static (shared) methods abstract (virtual/must-inherit). After debating this problem with Yaron, and a bunch of people on multiple programming forums, I decided to use the factory model and create instances of the classes instead of just calling static methods. After removing all static stuff from the classes, I added some abstract methods, that get called by formInputHTML(), which is now in the base class, and cannot be changed in the child’s, since it’s final. These abstract functions then get ‘filled up’ in the child classes and contain the service specific code.

I finished doing this yesterday, and am doing a very similar thing with the map classes in Maps now. The approach is basically identical, except for a change I made in how the map properties are handled (weakly typed instead of specificity defined class fields). This work is going hand in hand with another rafactoring to-do, which is the cleaning up of the Yahoo! Maps and Google Maps code in maps. In 0.1 a bunch of JS gets echo’d, and all the logic is done in PHP, which is a bad approach IMHO. I’m changing this to having one JS function to create a map, and just echo’ing the call to it with the correct parameters, and let the function do all the logic.

I expect the later do be done today, and when it is I’m going to add the centre parameter, which has been missing so far in Maps.

Closely following the initial release of Maps, the first version of Semantic Maps is now publicly available!

Some useful links:

• Semantic Maps documentation
• Semantic Maps demo’s
• Semantic Maps downloads
• Semantic Maps source code on the Wikimedai SVN
• Yaron’s (slightly more extensive) blog post about the release

Tomorrow I’ll be announcing both extensions. I can’t wait for the feedback on my work

Yes – The moment of the initial Maps release has finally arrived!

All Maps documentation has been updated, and now contains a download link, live demo’s have been added to my on-line dev wiki, and special sandbox pages have also been created.

No announcements about the release will be made untill Semantic Maps 0.1 is also publicly available, which can be expected to be tomorrow

Maps and Semantic Maps are both as good as ready for their first release!

I’ve completed all planned functionality for the initial release, and now only have to update the documentation, before publishing the extensions. I’m really happy the originally planed features are now completely finished. It took me quite some time to get all the remaining stuff finished, and worked on it practically non stop yesterday and today.

Yesterday I finished creating a new form input system and rewrote the Google Maps form input, like mentioned in my previous post. After that I started working on the one for Yahoo! Maps, which I finished today. Here you have a nice screenshot of a Yahoo! Maps form input:

I also finished the form input for Open Layers. This took quite some effort, mainly cause of the large amount of required JavaScript. The code to add a click event to an Open Layers map is in itself more then all the code needed to simply display a Google or Yahoo! Map, and is definitely a lot more complex. Anyway, I’m really pleased this is done and working nicely. Here you have another screenshot, showing an Open Layers map in a form input with NASA imagery as baselayer.

The last facet of the work I did was fixing up the display of properties in marker popups in Semantic Maps. I refactored the biggest part of the code responsible for this, and made some improvements to the actual output. This functionality is now working for all available mapping services.

The wiki code for displaying this map: {{#ask:[[Category:Locations]]|?Has coordinates|?residents|?capital of|format=googlemap|width=610|height=400|earth=on}}

The 1st Wiki-Conference New York will be held over the weekend of July 25-26 2009. There will be 2 days of events, held over a weekend in meeting rooms in Vanderbilt Hall at the New York University School of Law, and with activities starting each morning at 10 AM. Yaron Koren, SMW enthusiast and developer of …… a lot … extensions and my GSoC mentor, will be giving a talk about mapping with Semantic Mediawiki, including the upcoming extensions Maps and Semantic Maps I’m developing. By then both the extensions should have had their initial release, and I’m hoping to get some extra feedback and suggestions cause of this

Since this is a little far for me I can’t go though

I decided to really thoroughly rewrite the Google Maps form input (hook for Semantic Forms) in advance to staring on those using Yahoo! Maps and Open Layers. Before I started I made the prediction I would at least remove or move 125 lines of the 225 line long code. Want to know how many lines long the body of the current function is? No more then 26 lines! (And that’s spaces included.) A third of the code I just removed cause it was redundant, another third I referenced to the centralized mapping and Google Maps functions, and the remaining third I just moved to another location to make it centralized. After doing this, I added pretty much all functionality the parser functions for Google Maps have to the form input, which basically meant adapting the JavaScript functions, and creating some new ones.

Next to this refactoring and adding of new functionality, I also put the code in a separate class, SMGoogleMapsFormInput, which inherits from an abstract class, SMFormInput, containing all the common code for map based form inputs. This will make adding the inputs for the other services so much more easy Tomorrow I’ll attempt to get both the others working, and fix some small remaining issues, so the extensions can finally be released.

Here you have a nice screenshot of the Google Maps form input, showing a view of New York with the Google Earth map type.

The wiki code of the used form responsible for showing this map:

”’Coordinates:”’ {{{field|The coordinates|input type=googlemap|earth=on}}}

Last few days I made some really awesome progress on Maps and Semantic Maps. I fixed a whole variety of issues all over the extensions (to many to list here), and refactored quite some code to decrease code redundancy. A notable change to the exertions structures is the adding of a settings file, which contains a list of global variables that you can use to configure the default values for a variety of things, like the default zoom level for a map. I did this after a great suggestion of Sergey Chernyshev, author of the Google Geocoder extension, in a reply to an email I send to the SMW dev mailing list with the source code of my extensions.

There is still some work left before the first release, but I hope to have it finished by the end of this weekend. The most important things still on my to-do list are rewriting the Google Maps form input (hook for Semantic Forms), by letting it use the vastly improved and more centralized code since when I created it, creating form inputs for Yahoo! Maps and Open Layers and fixing an issue with the text in marker popups.