Why I like components…

This will hopefully be the first in a series of technical blogs dedicated to our in-house engine moFlow. The series will cover everything from engine architecture and resources to rendering and networking and will describe the technical challenges faced by the team and the solutions that were implemented.

Today’s sermon will provide a brief overview of moFlow’s architecture and discuss the merits and drawbacks of its component based design.

What is a moFlow?

moFlow (always a lowercase ‘m’) is Tag’s cross-platform C++ game engine; allowing 2D and 3D games to be developed on PC, iOS and Android with a series of social libraries for server-side and social network integration.

As with most names moFlow was chosen because it sounds cool and then a hastily designed reason had to be tacked on about flowing between mobile platforms. The spelling also used to change periodically until Bert traded the ‘w’ for something.

What are components?

Someone once described components to me as “nuggets of functionality” that are combined together to form game objects/entities. Essentially the behaviour of a game object is described by the components it contains and the object itself only has a transform. For instance a player object may consist of mesh, physics and 3rd person camera components. This approach is quite different to (arguably) the more common object oriented strategy; in which the player would have its own class (i.e.’is-a’ vs ‘has-a’ principal).

Why?

When designing moFlow the one clear goal I tried to keep in mind was modularity and that ‘plug and play’ mentality. I felt that this should extend to game objects and I was familiar with component based design from my experience using Unity3D. I really like the fact that an entity will always be an entity and never an object that derives from several base entities and that utilises some of (but not all of) its parents functionality (don’t get me wrong I like inheritance but only really as a tool to enforce behaviour and not to adapt it). Also if an entity is always an entity they are more easily pooled, as only a single pool is required.

I like that in order to customise an entity only a new component need be added and that the behaviour of an entity can change at run-time. This is incredibly flexible if, for instance; an entity needs to be selectable in the main game state but not so in a sub-menu. In that situation the ‘selection component’ simply needs to be disabled or removed. Similarly an object can become collidable simply by adding a physics component.

Kicking and Screaming

When I first started using Unity3D I found it difficult to get to grips with components. Some of the components uses were more obvious such as mesh component, etc. However I rarely created small logic components, preferring to create a new game object that controlled all the scene logic. In a sense making the structure of the program more like the traditional approach I was used to. When I came to implement components in moFlow I found that others suffered from the same issues.

Everyone wants to inherit from entity! It seems obvious to say but this largely defeats the purpose of the components. It’s easy to appreciate why developers are tempted to do this as the derived entity class can hold member pointers to all the components it contains and this does highlight one of the downsides of the component based architecture – getting hold of the components. In moFlow the components can be retrieved from the entity by name, index or templated type and the access is largely dependent on the container chosen. However I’ve always felt that this is a small price to pay for the flexibility that components provide. More frustratingly C++ does not provide a keyword that restricts inheritance and some people are so determined to inherit that not even the lack of a virtual destructor will stop them.

Another challenge is keeping components small and relevant while balancing granularity and genericism. Finding that balance really depends on the application but in moFlow some examples of components include sprite, mesh, light, physics and camera components and our applications have components for touch handling, meta-data and model-view-controller functionality. One of the key aims is to reduce component dependencies, this can often be achieved by communicating via the entity. One example of this may be a player object which contains a physics bounding volume and a static mesh. The physics component will update the entity’s transform and the static mesh component can either poll for the transform or attach itself as a listener for this change. If components do need to interact it is beneficial to use a weak ownership model; that way if one of the components is detached the other can adapt to this change (another method is to generate an event whenever a component is added or removed to which other components can listen).

Ownership

The component pattern often goes hand in hand with the factory pattern as it is far more convenient to create a method that adds and initialises all the components required by the entity; however ownership of components can cause issues. In moFlow components are often required by their corresponding systems (i.e. the physics component is updated by the physics system and the render components by the render system); as-well as the entity to which they are attached. Therefore I favour a shared ownership model that uses smart pointers to manage the lifetime of components.

Next Time…

I hope some of that was informative and I was able to get across how useful and flexible components can be. If I’m allowed back on the air again I will be talking about moFlow systems and resource management.

- Scott Downie, Lead Programmer

Yes Scott, you will most certainly be allowed back on the air after this great post and I for one am looking forward to the next instalment!

This announcement to notify all viewers of Tag-Cames.com that when you surf Tag-Games.com your computer will receive up to 21 cookies from Google Analytics. Do not worry, these cookies help us to develop Tag-Games.com into more appealing and informative website for you and are never given to any third party for any horrid advertisements (you will note that there are no, and will never be, any advertisements on Tag-Games.com).

This announcement comes as a result of a new European Union cookie law which obliges websites to inform and, where appropriate, obtain consent of viewers before cookies can be stored.

- The Tag Games Team

This week Kevin Black, our Lead Q/A test engineer discusses the challenges of testing dynamic service-model games:

Software as a service, not as a product

The biggest shift in project planning and milestone testing is the ethos “software as a service rather than a final product”, and this is reflected in the widespread approach of agile methodologies during both the development and live stages of new mobile projects. The old, almost “waterfall” approach of release milestones (Alpha, Beta, Gold Master) is becoming eroded and Tag’s QA has to adapt accordingly.

My personal test career started in AAA development for the current generation of consoles – both XBLA downloadable and full retail – and whilst agile methodologies were used in the development of those games, they were primarily used for manoeuvrability when delivering a piece of software to a publisher. This meant that agile testing was limited to the delivery of a milestone: a first playable build, a content complete alpha build, various beta test builds, a final release candidate gold master build, etc. This is no longer the case.

Brave new devices

Mobile devices are now major gaming platforms and a wealth of dynamic content is now available to mobile users. It is easier to update and improve software after release than it has ever been before and this puts our customers first, allowing us to develop the application in real time, taking into account user feedback and feature requests.

However, this also presents new challenges for test engineers in the QA department and breaks the typical flow of Alpha, Beta, Gold Master release milestones. It is now a question of getting the product into the customer’s hands as soon as possible so as to generate feedback and shift the course of the service in accordance to requests. While the first release is still given the same test approach as a console final release candidate, we get the product out quicker by hitting a minimal viable product build which has the same stability of a gold master but only some of the polish.

How Tag copes

Test case scripts are still used for various milestones and rigorously carried out on any new update, but during the post-release development phase – when player feedback is driving design modification – our testing approach becomes more “ad hoc”. Possible design changes to the game are coming thick and fast and they need to be kept on top of, some are experimental changes, and some are sizable reworks of old features. It all needs to be reported on immediately.

When a new feature-set is developed and ready for update submission however, the focus changes. Each update is essentially a new Beta version of the game that requires rigorous testing before delivered to the customer. Test case / white box scripts are used to confirm the expected behaviour of every facet of the game in addition to the utilisation of more ‘open’ testing techniques – everyone here at Tag is playing our own games all of the time and feeding back to the test team, for example.

Organic and alive

These are the massive advantages to the new mobile platforms. They offer flexibility like we have never seen before and give us the opportunity to ensure we are delivering the best possible games. We must remember however that it is no longer the case that testing for a game is finished when it is boxed and shelved. Software development is now organic and alive, so our testing must become organic and alive too.

Over the past 5 years Joe has worked on many exciting projects including Car Jack Streets, Doctor Who: The Mazes of Time and Football Pro, and has had the opportunity to watch the studio grow from its humble beginnings of three people working in a shed to being a fully fledged mid-sized mobile software development studio with international clients and industry-leading technologies. Without Joe and his coding talents then, the studio simply would not be where it is today, so we all thank him for his tireless work! (We are sure Joe’s lengthy stay at Tag is due to the studio’s great working environment and hilarious people, and not the awesome Tag placard and retro gaming presents he received!)

Given the nature of Joe’s milestone we decided to ask him a few questions to discover just what he loves about Tag and his work:

So Joe, why did you join Tag Games?

Joe: I applied for the job at Tag because they seemed to be making different sorts of games to everyone else. That’s been one of the good things about working here, I’ve got to work on games in a lot of different genres, and the shorter development cycle of mobile games has meant I’ve never on the same project for too long. In another part of the industry I might still be on the same project now as when I came in the door in 2007!  I’ve been given a lot of responsibility right from the start too, which has been tough at times but it has allowed me to be a lot more creative, and I’ve been able to work on all aspects of each game, rather than just the AI or menus or something.

How has Tag developed then since you first joined the studio? Things must be completely different!

Joe: It’s changed a lot since 2007, I think there were only 5 people in the company by the end of that year. The number has varied a lot since then, with new people joining and interns coming and going, although there are still a few of us left from that first year. In terms of the development process I’d say we are much better organised and streamlined now, whereas back in the day we had to improvise. It was inventive, and back then it worked okay for us on those small projects, but obviously things have to become more structured if you want to grow, and that’s what we’ve done.

Which have been your favourite projects?

Joe: The project I’ve enjoyed working on the most so far has been Doctor Who for the iOS devices. It was a bit of a coup when we got that contract from BBC Worldwide, and they were a really good client to work for. It’s also the piece of work that I’m most proud of, probably because I coded it almost entirely from scratch. Every project since Doctor Who has used our in-house engine moFlow, so in some ways it represents the end of a particular era at Tag for me. I’m also pleased about how well Car Jack Streets has done for the company, and I’m sure the sequel is set to do even better when it arrives.

Finally, what has kept you at Tag for so long?

Joe: The thing I like most about working at Tag today is the great staff and working environment we have. It’s a fun place to be, we have a laugh and there’s always a good team spirit even when things are tough. I think that’s why I’ve stayed, it would be the hardest thing to replace.

So there you have it – Tag Games, it turns out, (like we didn’t already know!), is a great place to be! Hopefully we will have more employees reaching their fifth year milestones soon, but until then, we’d all like to say a resounding congratulations again to Joe!

Firstly, Innovation cannot be confused with invention.  Invention is commonly described as “The creation or design of something that has not existed before”.  In contrast innovation is described as “making changes in something that is established “.  It is a key differentiation and one that needs to be clearer in the gaming industry and will be the source of countless discussions well into the future.

The role of data analyst

The general perception within the gaming industry is that analysts sit and pour over data looking for ways to maximize revenue from users.   That we are much in the same mould as marketers and see a game as a series of numbers or metric events and that user fun comes second to revenue.  Our role is further compounded by companies like Zynga and Crowdstar who return extraordinary revenue returns and glorify the use of data analysis and have led to the concepts of “data-driven” design and “big-data” now common place.

It is true that in the F2P model a data analyst has a responsibility to make suggestions that will improve a game in order to maximize profit, retention or engagement.  To suggest, as Huztech’s blog implied, that data analysis won’t lead to innovations in gaming is canvassing all analysts with the same brush and is far from the truth.

The use of metrics is not a new practice in game design nor does it stifle a developer’s creativity. The use of heat maps to visualise popular routes in FPS levels or in usability testing such as Microsoft’s Happy Action Theatre are perfect examples of where metrics have other uses besides revenue.

First Hand Experiences

At Tag we identified through the use of data a significant proportion of users leaving Funpark Friends early before the end of the tutorial.  Without substantial soft data through forums, emails and other social media identifying where the issues were, fixing any problems would be extremely difficult.

The use of hard metric data allowed the team and myself to sit down, pinpoint where users dropped off and gave the designers an ability to focus specifically on improving key aspects of the tutorial ultimately making the game more accessible to a new gamer.

Recent experiences with Funpark Friends allowed us to see how important the different types of Rides were to user’s enjoyment.  Gamers were eager to progress and see what new and crazy rides unlocked at each level.   The data showed the positive impact that rides had on progression and gave the design team even more confidence to be more creative and come up with even crazier and amazing things.

It is this objective data that when used in combination with other information sources that risk can be reduced and accurate informed decisions can be made.

Risk in Gaming

Game design and invention is inherently driven by risk.  Few companies in the current climate can afford to spend millions of pounds on a project that may not have any success.  Instead what we see now is an industry that views iteration as safer than invention.  It’s no surprise to see sequels dominate the console market and iterations of successful games flood both the mobile and Facebook markets.   You only have to look at the recent mainstream stories surrounding Nimblebit and Buffalo studio’s to see this.

Data analysis allows for objective analysis to be carried out aiding the development of a game and ultimately improving the product for the user.  Reporting back to the various team key data and trends allows decisions to be made with maximum confidence and minimal risk.  At Tag especially they see the use of data analysis as another tool to compliment the work of the designers/artists and programmers.  In the current climate a company needs all the tools it can at its disposal.

Data Driven Design vs Data Supported Design

Ideas will always come from talented individuals but where to concentrate these ideas is subjective and brings about risk.  The use of metrics now allows studios to take more risks, be more creative and to ultimately be more inventive.

Is it a surprise that in most cases the companies that see revenue as being more important that invention is least apparent?  Are both mutually exclusive?  At Tag our aim is to do both.  Be inventive, creative and revolutionary with our ideas and still generate the revenues that a successful game should do.

So while data will not provide the next big thing in gaming, the use of data will be key in the development of the next big thing and surely that must be seen as being innovative?  Where that next big comes from we don’t think we have far to look.

The moFlow social mobile games engine, developed in parallel with ‘moMetrics’, a complete data analysis solution, and ‘moConnect’, a complete scalable server and player-account infrastructure solution, allows us to develop 2D and 3D player-to-player applications for iOS and Android devices simultaneously. One of the first of its kind, mo technology provides a multi-platform social mobile development package with support for web, Windows, Mac and other hand-held platforms in the works.

Funpark Friends for Android (released under our ‘Tagplay’ brand) is scheduled for launch early February and will allow players to log-in and play using their cloud-stored game data on any Android or Apple device. What’s more, they will be able to interact with other Funpark Friends players regardless of the devices being used. Our talented Scott Downie, architect of the moFlow engine, outlined how mo technology makes this possible:

“moFlow’s modular design allows functionality to be added or removed dependent upon the specific needs of the game.  Application sizes are therefore much smaller than those built with other engines, whilst providing support for a variety of proprietary and 3^rd party libraries (including our own moConnect and moMetrics).

Our scalable code solution allows moFlow to take full advantage of multi-core hardware – benefiting current and future platforms without any additional work – and its powerful UI system enables users to customise look and feel across a variety of screen sizes, resolutions and languages.”

When recognising the growing importance and market share of the Android platform we found ourselves, like many independent studios, unable to fully capitalise on the opportunity due to the increasing cost of porting code from the lead iOS platform. moFlow eradicates this problem, allowing us not only the ability to deliver social games which are device agnostic but also launch content on the two largest App markets simultaneously.

Paul Farley, CEO commented:

“With the launch of Funpark Friends on Android using our proprietary moFlow, moConnect and moMetrics technology stack, we have demonstrated not only our continued excellence in game design, but also a technical capability that puts Tag on par, or ahead of the competition. We now look forward to extending platform support beyond mobile and exploring how we can use this technical advantage to help other independent mobile developers deliver better social games, across more markets and at a lower opportunity cost.”

Well, you do now!

Tag’s In-House Data Scientist

As the resident data scientist at Tag it is my job to get to the heart and soul of what makes a good game and, more importantly, how to make it better.  I look at what you the players like to do in our games and make sure you can do more of it. I also, again perhaps more importantly, look at what you don’t like to do and ascertain why that is the case so that I may best advise our designers and artists of how to rectify issues in the game’s mechanics which are often at the root of your dissatisfaction. What is cool to note is that this approach to game development is so new that no one else in Scotland is applying data analysis in this way – Tag, it seems, is again leading the curve as far as mobile games development is concerned!

Data Analysis in Games

That is not to say that metrics and analytics are new concepts for games developers however – companies like Zynga and Playfish see these facilities as key in the development of their Facebook games – but it is only now, with the advent of freemium applications, that we are seeing these facilities becoming a core component of the more expanded mobile games industry.  Terminology like retention, DAU’s, MAU’s and conversion rates are steadily becoming common place across the mobile games industry and there is a clamour to understand this data as best as possible and as quickly as possible. You only have to look at the success of Zynga and Playfish to understand the effect data analysis can have and why it is becoming as important as it is.

Mobile 2.0

In order for us to be successful in the future then we have to embrace Mobile 2.0 gaming within the free-to-play model.  It’s not enough to put a game out for free digital distribution and watch the sales roll in anymore – the service model is all about two way engagement. We want to make the best games possible, but it’s only by listening to what you say and looking at what you do that we are able to make the changes to the game that you want to see.

The Future

What I hope you have taken from this blog is that we at Tag understand that all of our users are key to the on-going development of Funpark Friends and to the development of our future titles.  Through playing our games you help to develop our games, so keep playing and thank you!

Chris

Like our Facebook Page - http://www.facebook.com/Tagplay

Follow us on Twitter – @taggames and @tagplay

We had just moved into our new office and had received news that the first social mobile game to be released by Tagplay – the self-publishing sister brand of Tag Games – had passed the Apple review process and was clear to launch on the 1st of October.

Well, here I sit, a mere 25 days later from a successful weekend launch and I am awed by how far we have come, how much we have learned as a studio, and how long ago the launch feels despite only being 25 days ago! So, whilst there have been many awesome new additions to the office that I would love to tell you about – like our new foosball table, our new bean bag brainstorming area, or our new remote controlled flying fishes (yes, you read correctly… and yes, you should be jealous) – this blog update will be dominated by news of the launch of ‘Funpark Friends’, the first social mobile game by Tagplay!

On September the 31st at 11:59pm we took a deep breath, flicked the button on our Apple Account and released Funpark Friends onto Apple’s App Store.

It was dramatic, but to both our joy and relief we had our first customer transaction just one and a half hours later. Truly, that first transaction was an exceptional moment for Tagplay as, despite only being for the sum of £0.69, it represented 9 months and thousands of hours of work coming to fruition. We had released a gorgeous game, built on our very own moFlow engine, supported by our very own server team and which was sending data back to us on how our users were playing the game, allowing us to treat Funpark Friends as it was always intended to be treated – as a service, not as a product. We had done something only a handful of studios in the country had done and we were overjoyed. Funpark Friends was online, it was being played, it was being loved and it was making money.

Of course, the launch of Funpark Friends was not all sunshine and roses. A bump on the road came when we realised that demand for the game was far greater than had been anticipated. The result was that the Funpark Friends server structure became overloaded and many people who had installed the game were simply unable to play the game. This was a huge blow and many potential Funpark Friends players were lost forever. Within the space of a few hours the server structure was expanded to handle the demand, but the damage had been done and a sobering lesson had been taught.

However, despite launching Funpark Friends in a soft manner, and despite the server-demand hiccup, Funpark Friends has had fantastic DAU and retention rates since it was launched a little over three weeks ago. We can only be extremely positive taking Funpark Friends forward into the future then as large return-on-investment is possible from extended market pushes if monetisation of the game is optimised. Certainly, analysis of how our users have interacted with the game thus far has raised a number of issues which can, and are, being addressed in real-time. The capability to react in an informed and speedy manner to such issues will allow us to improve the playing experience of Funpark Friends over time and improve already very palatable customer acquisition and monetisation figures as a result. As previously mentioned, Funpark Friends is service and like all services it must be updated, improved and expanded if it is to be successful.

Our commitment to developing Funpark Friends as a service was exemplified just yesterday when a cut-down HTML5 version of Funpark Friends was released in partnership with the leading mobile social network MocoSpace. This version contains the core gameplay dynamics of the native game and is the first step in bringing the full Funpark Friends experience to web browsers. Additionally, an update of the native iOS Funpark Friends app has been implemented and another content update is due within a few weeks. Finally a HD iPad version of the full game is expected by the end of November.

All of these developments are being conducted with the aim of creating one cross-platform, social, mobile service that will satisfy the demand of our user base for a fantastic casual gaming experience.

There is plenty of work ahead then and plenty to do before Funpark Friends can be the success that it deserves to be. I can only imagine that in another 25 days I will look back on writing this blog update and be in awe again of how much we have learned, how far we have come and how long ago today seems.

Simon

Click here to get the fun and excitement of Funpark Friends on your iPod or iPhone now!

With our new office we have doubled floor space and everyone is appreciating the extra room. We installed table football today, but despite the demands from certain team members (I’m looking at you Angus) we’ll be stopping short of the 10 pin bowling lane! The local media have been quick to pick up on our plans to increase the team size from just under 30 to around 50. It’s great to be in the spotlight for a few days and it’s satisfying to be adding more positive news stories to Dundee’s gaming history. I hope there are many more to come as we grow to become the #1 indie mobile games company in the world.

Before

After

Thankfully all the major work in moving office has now been completed and this very evening we recieved an email from Apple to say that Funpark Friends has passed review and is clear for launch on Saturday 1st October. Phew!

It’s time to wipe the sweat from our collective brows and get ready for the next chapter in our social mobile gaming adventure. The next couple of weeks are going to see the culmination of over 9 months of hard work and we just can’t wait to share Funpark Friends with you! Watch this week for a game-play trailer and more details on the game and then from next week we have some very special promotions for you to get involved in. Here’s a sneak peak of the special edition T-shirts we’ll be giving away. The prizes only get better from here on in!

Funpark Friends T-Shirts

Hopefully in the next update I’ll be able to share some facts and figures from the Funpark Friends launch. Until then make sure you’re following @Tagplay on Twitter and are watching the Tagplay Facebook page for details on the game and how you can win one of these amazing T-shirts!

P