Hourly Fragments

I thought this would be pretty tough because I needed change the way scripts worked. Initially, a building had a unit generator script, which obviously created the units, but I needed a few things. To 1) be able to assign different sprites to different units. 2) assign different weapon types 3) assign different health values

I also need to have the buildings pump out units at different timing intervals, depending on how powerful the unit might be. (stronger units take longer to produce but obviously have more impact in combat).

So instead of having all this detail in the unit generator script, I really need to break it out into classes to make things more contained and maintainable. 

Something like a class for factories, which can produce a squad (class), and a squad is made of units (another class), which will have the aforementioned stats and Unity gamey things like colliders etc.

I could technically create unique functions for each unit type but really this is duplication of effort when they'll all have the same basic properties.

It took less time than I thought. Have a gif!

I wanted to do a couple of things with my next update. 1) I wanted to have a group of units produced at once 2) I wanted to put them in groups

I got there after a couple of hours

I know I shouldn’t have, but I got distracted with another side project. I mean, apart from general family life that’s basically meant I haven’t posted here more than an entire year.

Anyway I have another mystery project underway. I basically wanted to start fresh and see if I could just work it out myself.

One of my favourite RTS from the early days of gaming was Z. It's possibly because it was probably more a real-time-tactical; no resources gathering, no manual building of units, simply send your units out to capture factories that produce more units at certain intervals, then send those units out to attack enemy forces and capture enemy bases.

The whole mechanic leverages around capturing as much as you can as fast as you can to get units pumped out of the factories as quickly as possible, while also trying not to overextend, leaving your territories open to capture and having their production turned over to your enemy.

I really enjoyed it and wanted to create something inspired by the experience because it's something I want to play.

Here's a building pumping out units on a timer.

Game done. That distraction I mentioned? It’s done. I mean, as far as I can tell, it’s done. There are probably some enhancements I could make, and there’s probably a bunch of broken stuff. Maybe one day I’ll tackle the enhancements or fix the broken things, but probably not, unless it’s REALLY broken, or I’m SUPER bored.

You can play the “finished” product at the link below.

http://textadventures.co.uk/games/view/nvifhoflnkmax2-hhojnkq/uncharted-4-5

Thanks Tim Colwill for an amazing script, thanks for not complaining that I appropriated it for this stupid thing.

Thanks to textadventures.co.uk for your text adventure engine. I think I mostly did what I wanted to do, mostly in ways I wanted to do it, and you made it pretty easy.

Thanks anyone who has played this, or plays it now, 6 months after I started it or whatever. It’s been an experience, to be sure, and I’m glad for it, but also glad to be done.

One of the important systems for this game is player location on death. In my previous post message received, I talked about how I got this working with the new networking system in Unity. 

What was missing from this was the actual player coordinates; I was just sending a hard coded string. I had a bit of time to mess around with it and decided I’d get some real data being sent. Turns out it was pretty easy!

Instead of setting  msg.deathCoords = “0,0”, I needed to have it send the transform position coordinates of the player, specifically, just the x and y values, and I needed them as a string. Well, ToString() worked nicely, but I also didn’t need a float value to eight decimal places or whatever, so I’m not sure if it’s the “best” way, but typecasting the transform position as an int did the trick here.

So in the screenshot above, you can see I’m recording the server timestamp, and publishing a static “player died at coords” message. In order to have the actual coordinates published at the end of that, all I needed to do was change:

msg.deathCoords = “0,0” 

to

msg.deathCoords = ((int)transform.position.x).ToString() + ", " + ((int)transform.position.y).ToString();

Simple, no?

It didn’t take much work, didn’t cause me any problems, and was quite easy to implement. But I’m fairly pleased with the outcome because I’ve achieved a specific goal I set for a session of dev work, and it’s one less thing I have to worry about later. Unless I break it.

It’s been a long time since I’ve updated this, and indeed since I’ve done any real work on the game. One reason for that is a bit of choice paralysis. I need to implement a tile system so I can make maps/levels. I can use Tiled, and a Unity plugin, or I can write my own, or I can wait, because it turns out Unity is currently has a tile engine in development and it looks pretty cool.

So I’m trying to decide which way I should go.

The other reason is I started working on another game based on a fake script for an Uncharted movie I thought read a lot like a text adventure. It was a stupid thing to do, and even though it’s entirely ridiculous and should never be played by anybody, it’s nearly finished.

You can find it and a little more info at the link below.

http://textadventures.co.uk/games/view/nvifhoflnkmax2-hhojnkq/uncharted-4-5

One thing definitely need to doOH WAIT, I solved the problem with my scroll view not updating! It was all my fault! I am a buffoon! It was basically that I was updating the text to be just the initial starting string every update/frame. So yeah.... where was I?

Right. One thing I definitely need to do for this game is work on my tile engine. Isometric tiles can be a little tricky to wrap your head around, but I have a working tile engine coded, and it basically works by storing the tiles in an array and then I call an element of the array to place my tile. 

The problem is this is all done in code, and I absolutely don’t want to design my world in code, so I need a level editor, but I absolutely don’t want to code a level editor if I don’t have to. And I don’t.

It turns out there’s a pretty neat tool called Tiled and after a quick play, I’m fairly impressed how powerful it is. Import a tile sheet, set up your layers, then just click until you’ve built what you want.

The saved *.tmx Tiled file is basically an XML with CSV, so my task is to work out how to read the tmx and render it with Unity - not a huge leap from where I am at the moment. Of course this is going to get a lot more complicated once map complexity increases. I’ll have multiple layers and heights, understand how and where the player can walk behind/in front of objects, where they might collide or be unable to pass, and probably a bunch of stuff I haven’t thought of yet.

Now some of this I’ve done previously when I was messing around with a tile engine in XNA (before Microsoft decided to axe it), but that was also code-based. It was giving properties to certain type of tile to say, for example, whether it was traversable or not. And that’s fine. I may end up doing that again. But I I may also be able to make this more user friendly for myself with something which Unity calls a “prefab”.

I don’t actually know if this is something that’s possible, but it seems like I should be able to assign a bunch of properties to a gathered set of tiles that make, say, a tree, and have the prefab be like: yo, this is a tree, a player can’t walk into the trunk, but they can walk around it. And then every time I want to place a tree, I use that prefab and it always has those properties.

Maybe? I don’t know. I can do this in code of course, but if I wanted to just code everything from scratch I wouldn’t be using an engine like Unity.

So I guess this is a long winded way of saying I didn’t achieve a whole lot since my last update, and I have heaps of research to do to progress down my current trajectory. Sorry I made you read it.

There are no words for how good it feels to get something working after such a long time. My hourly dev effort has definitely not worked out the way I wanted since I hit this problem, but I now have a solution to a couple of problems.

The first was to replicate an Unity’s old RPC system in Unet. RPCs still exist in a sense, but they function differently in the HLAPI, and have weird restrictions which means they aren’t suitable for what I want to do. In this case, send a piece of information, say, a simple string, like “I really don’t like Unet”.

The way to do this is actually with Unet’s network messages which you can read about here: http://docs.unity3d.com/Manual/UNetMessages.html

And if that doesn’t seem to make a whole lot of sense, here’s a fairly simple breakdown that got me started: http://forum.unity3d.com/threads/command-between-2-different-scripts.334902/#post-2168275 The forum post (Thanks Carpe Denius!) gave me the basic approach and understanding, and going back to the official documentation helped me flesh out what I needed to do. 

Using the MessageBase class it’s possible to create a message with a range of data inside, but for now I’m just sending the aforementioned simple string.

On the client, in the one script I have

class StringMessage : MessageBase {    public string deathCoords; }

and

public class Networking : MonoBehaviour {    NetworkClient myClient;    StringMessage msg;

// Use this for initialization void Start ()    {        ConnectToServer();    } // Update is called once per frame void Update ()    {    if(Input.GetKeyDown(KeyCode.P))        {            Debug.Log("Pressed P");            msg = new StringMessage();            msg.deathCoords = "0,0";            myClient.Send(666, msg);

       }     }

//some other functions doing some stuff... }

So you can see I’m going to send coordinates of where the player has died (simulated by pressing P), and for now those are hard coded to a string saying 0,0. I’ll retrieve the actual player coords later, I just needed to get the message sent first; baby steps.

Then on the server

public class MyNetworkManager : MonoBehaviour {    public StringMessage receivedMsg;    public bool isAtStartup = true;

   void Start()    {         SetupServer();    }

   void OnGUI()    {         //some display stuff    }

   // Create a server and listen on a port    public void SetupServer()    {        NetworkServer.Listen(4445);        NetworkServer.RegisterHandler(666, DeathLog);        isAtStartup = false;    }

   void DeathLog(NetworkMessage msg)    {        receivedMsg = msg.ReadMessage<StringMessage>();        Debug.Log("received Message " + receivedMsg.value);        guiUpdate += "player died at coords " + receivedMsg.value + "\n";    } }

And here I’m logging the location. one of the things I was also doing wrong here in the DeathLog function was not setting a local stringmessage variable, and trying to do this

Debug.Log("received Message " + msg.ReadMessage<StringMessage>() .value);  guiUpdate += "player died at coords " + msg.ReadMessage<StringMessage>() .value + "\n";

This gives an error because you can’t deserialise the message more than once, which happens when ReadMessage is called. So by deserialising into receivedMsg it’s usable as many times as required. I spent a few hours trying to work out why the log would print out fine and then every other readmessage afterwards would fail.

I found that solution here: http://forum.unity3d.com/threads/need-help-with-custom-messages.339010/

So now I am sending and receiving messages. The next step is to fix my server GUI, because despite me updating it, the display does not change. I’m hoping that’s less painful than these last few weeks.

It seems unnecessarily difficult to do some really simple things in Unity’s Unet. Where RPCs in the previous Unity3D networking made getting information from a client to a server fairly easy, UNet instead wants to make it incredibly difficult. 

Or maybe it doesn’t. It’s actually impossible to know because one thing that Unity is absolutely terrible at is tutorials. The documentation is pretty awful, and doesn’t explain anything very well. The scripting API is equally barebones, showing code snippets in the rawest form of their implementation. All I want to is send some text from the client to the server. You’d think that would be pretty easy.

If you want to code something in Unity, it’s like you need to already know how to do it, which is pretty counter-intuitive, unless you suspect as I do, that it’s so Unity can sell support.

I’ve been beating my head against a brick wall of networking nonsense for four hours and made roughly no progress whatsoever. I’ve been reading the community pages for anything related to Unet, ClientRPC, or Command, as well as watching various video tutorials, and found nothing that helps; so after the connection win last week, I’m back to posting for assistance. And wondering how much “support” would cost.

Sigh.

Great success! I have a client and server connection! I mean, for some reason it says there are two clients connected and I know there is only one, but baby steps. I’ll fix that later.

I’m particularly excited because this is working on my desktop, something that with the previous networking implementation, did not happen. I spent a good long while tearing my hair out, Googling, searching community forums, and documentation to work out why my client and server wouldn’t connect on localhost. It wasn’t until I ran on my laptop that I found my code was fine.

I’m now fairly sure it is the fault of my Bigfoot Killer Networks LAN chip on my motherboard - this popped up as a problem trying to play an online game, and a swift netsh winsock reset dumps the LSPs and allows connectivity. Of course it might be that the new networking system in Unity 5 fixed something else that made it work on my desktop... but probably not.

So I have networking. Next step is implementing messages to track player death.

But a quick note on Unity Client/Server setup. This new “UNet” offers a High Level API to facilitate simple game client/host setups, for example if your game was multiplayer and no hosts were found, your client would make itself a host. 

Neat, but not at all what I wanted. And if like me you want a dedicated server and clients incapable of turning themselves into hosts, you probably want to look at this section of the documentation: http://docs.unity3d.com/Manual/UNetClientServer.html

Yep, it has both client and server functionality in the one class, but it’s easy enough to parcel out into separate applications. Okay? Okay.

The next bit might be a bit boring, but like SVN, setting up some kind of project plan is mandatory for you to know what you’re doing. So I signed up to Trello.

Sneak peek at some upcoming features!

Trello is used a lot by agile projects, and makes it pretty easy to type up tasks, and drag them around into various columns as you work through. It’s also good because if you’re working and something fires off a mental note to do something else, add new functionality, whatever, it’s really easy to type up a card and throw it in your backlog straight away so you don’t forget. 

You can also assign verbose descriptions, and checklists, and assign people to tasks if you’re working in a team. I like it.

I already have a design document, so I know what to do, at high level, and sure, I could work from that doc directly, but this makes it a lot easier to break down all those high level tasks into things that are more manageable, and track what I’m working on and see where I’m at; I can’t help feeling (like with version control) that these “boring” bits will save you a lot of time and hassle later on.

Finally, something I’m going to start doing, for ease of traceability, is keep a running tally of development hours spent since I started this blog, and estimating an additional 50 hours of work prior.

I actually used to have networking...working. I was sending and receiving messages from client to server (and vice versa), and then Unity went and changed the way their entire networking system functioned when they upgraded to version 5. 

Everything that I had previously done was deprecated, and all of a sudden I had to learn a new networking API. I still have to, I suppose. In about two hours today I managed to resurrect the old server side GUI, and rebuild the dedicated server code - at least, rebuild the initialisation of the server, as I haven’t yet had time to look in to how to replace the RPCs which were sending/receiving my messages.

Once that happens I’ll need to check that the server log is updating with the required information, and that the scroll pane is handling it. DEER LORD that scroll pane. Out of everything I’ve worked through so far, finding a scroll pane solution that did what I wanted was by far the hardest. Hopefully that hasn’t changed in Unity 5.

It feels like Unity’s 2D animation control is quite good. I don’t have a lot of experience with other tools, mind you; Unreal didn’t have a native 2D mode back when I started this, I’d messed around with GameMaker having bought it in a Steam sale, and had actually worked on some animation and tile engine stuff in XNA, right before Microsoft announced it was no longer supporting XNA - but Unity was popular, had native 2D, and as far as programming goes, .NET is just nice.

So today’s fragment was spent developing my half-done character animation system. I’m basically binding player movement to directional run and idle animations. So using WSAD the player runs up down left right (or north south east west) and when they release WSAD, they stop running and stand idle facing the last run direction.

Pretty simple except I got my state flow wrong the first time around and am now fixing it.

I’m also having some trouble with the best way to run to the diagonal points, from a coding perspective. Using Unity’s Input.GetKey works for one key, but not so well for multiple keys, so I need to find a way around that.

There’s also the issue that state transitions don’t seem to work, and I’m pretty sure these are exclusive to the idle animations transitioning to run animations. See from any state, it’s possible to specify that the animation doesn’t have to complete, e.g. if I press and release D, it’ll start the run east animation, but if it’ls only played two frames of the eight in the animation loop, it’ll will stop the animation.

But since I’m not defining any transition from idle to anything else, I don’t seem to have the option to do this, and the animation continues playing to the end; so if I release D then quickly press S, the character moves south, but is stuck in the east idle animation doing so.

The solution might be in code rather than GUI, and it might take some investigation to solve this particular issue.

In the meantime, I have a player character animated, running from any state!

One thing that you should definitely do with any big computer-based project is create backups. Probably the best way for this to occur with a project that involves code, is by using version control.

Version control, in case you weren’t aware, allows you to keep file backups, but also keeps revision history of those files. So if for example you made a change that completely breaks your code/app/game, and you aren’t able to change it back (because maybe you forgot how or what you did) it’s possible to restore from an earlier version. 

If you also happen to be working in a team (I’m not) and multiple people change the same files, version control allows you to merge the changes before committing them back into the project.

For some time now I’ve been using Dropbox, and that’s been acceptable; it has some rudimentary file versioning, and it’s cloud-based so I can sync projects between PCs easily enough.

But something happened a while back where I upgraded Unity3D, and my game project no longer worked, and I was unable to restore a working state. So today’s hour I chose to spend installing and configuring SVN or “Subversion”, an open source version control system, to ensure my dev work from here on out is a little more protected from loss.

Aside from SVN, I’ve used both CVS and Github previously. I found Git to be fairly terrible, and CVS I didn’t mind, however I also happen to have a Synology NAS which has a SVN server app on it, so the choice was pretty simple.

Along with TortoiseSVN which provides a relatively straightforward GUI to SVN controls, it’s a good solution. The NAS has redundancy mirroring, and the project folder also still sits within my Dropbox directory, so I know everything is backed up.

I read somewhere that Confucius said: humility is the foundation of all virtues. 

But rather than a single solitary hour, let’s say it’s an hour day. Something achievable for the time-poor. Something that one could, if one wanted, make time for. 

That’s the idea anyway. I’ve been working on a game for about a year. Not my game; it was designed by someone else, and at some point I may reveal the concept and design. Perhaps I won’t be allowed to. Perhaps it’ll become something larger. I’ll have to wait and see.

So it exists, or existed, out there on the Internet, and it looked interesting; a group of systems I thought would be interesting and fun to implement, and I wanted to learn Unity3D.

I stumbled forward and to date, after a year, I’ve spent a grand total of maybe 50 hours on it, and that’s a small showing of effort over such an expanse of time.

The hope is that by documenting my experience, problems, solutions, I’ll have additional impetus to continue. Something to show that isn’t a bunch of build executables in a directory hidden in My Documents, and a reference point for my learning.

My aim is to set this as a target, to try for an hour of game development work a day, seven hours a week. It could be research, animation, tutorials, whatever helps further the project end goal. It’s a soft target, an effort to try, by necessity, so if I’m unable to find or make time one day, it’s not failure, it’s not a setback, it’s simply something I will try again to achieve the next day. Maybe I’ll be out one night after work, but maybe I’ll have five hours uninterrupted on the weekend.

And this time next year I hope to lift my head and look around, to take stock. I don’t expect a finished product, but it would be nice to see something that looks, if you squint just right, like a game.