Vapourware Codexian Game Development Thread

[what am i doing]

I think i'm hogging up the thread a bit :-P.

That's on us for not making progress.

EDIT: what little progress I'm making: reworking my editor so that instead of an ImGui window with buttons that open up more windows I have kind of an Adobe style layout and can pick the thing I want to be editing on the left, I get some data / options on the top if applicable, and more data on the right. This will take a while to shift around because it means using std::function to configure the UI and somewhat changes the way that the UI is described. But it will make it much easier to use eventually and will also make it easier to link active brush modes with the selection of an editing mode.

Now because I'm still kind of burning out on code I have been doing some world design stuff as well. Part of the trade simulation will require the presence of defined geographic regions in order to batch together very small trading units (such as one-tile-minor countries) and to subdivide very large trading units (such as late-game pan-national states or major trade blocs), for multithreading purposes. I tried just picking clusters of zones and this did not work out well. So instead I am going over the rough draft of cities I had and redoing them, with the cities roughly equally distributed (except that major rivers will have cities at their mouth instead of somewhere vaguely nearby, and where a city would be near the course of a major river it's at the river instead). I am then periodically generating a Voronoi diagram out of the cities to verify that the resulting areas will be roughly equal in size, overall. Once that is done I will do a rough sketch over the final diagram, and then trace along the closest zone borders to produce trade regions.

The idea here is that trade between populations and operations takes place per Zone first, but there's also an entity called a port (not necessarily an actual sea port, but generally linked to the major city in the area) which can purchase and sell resources. The port is accessible in that scope, but then in the next stage, the ports from various zones in a geographical region are able to trade with each other, plus one additional regional port (or possibly two, one to represent the major trade city in the region, and another to represent "diffuse trade"). Then the regional ports trade with each other plus one global port. Then the global ports trade with each other. To my mind this should provide a reasonably accurate simulation of the flow of goods in a global market while also allowing aggressive parallelization - but (eventually) we'll see how well it works in practice. In any case, that's why the trade regions need to be sorted out. I do want the simulation to be able to scale very greatly as average core counts increase, so rather than having a dedicated worker thread for each category of work (as Paradox does in their newer games, and which would at a certain point lead to idle cores) I prefer each category of work to be able to be sliced apart as much as possible and fed to a thread pool. So far as I can tell, we still haven't run into a "core limit" - AMD is putting out CPUs with 192 logical cores, and while those kinds of core counts won't be in average consumer hands for a very long time they do prove that there's no physical limitation against having a couple hundred logical cores (unlike clock speeds, which have basically hit their limit).

 

[barker_s]

Reminds me I haven't posted here in a while either. Haven't had much time to do gamedev since the birth of my third kid, but the dust is slowly settling down.

I am still working on my game framework, and I thought that it would be a good idea to start the dogfooding process. So, I started working on a small project that would help me test the features of the framework in question. Decided on a multiplayer tactical dungeon crawler of sorts. It's unlikely that I'll ever finish this one. Its main purpose was to help me find bugs and assess certain decisions from the usability standpoint.
Here's the little video of that project:


So, what now? As I said, I had a little break from coding, so when I went back and took a look at the framework's code I decided to... rewrite it . Not all of it, mind you, but with a fresh set of eyes, it became obvious to me that certain parts of it have become Byzantine for no good reason. The biggest offender is the C# source generator module. In principle it functions well, generating a lot of boilerplate and relieving the programmer, but given the amount of code it outputs (e.g. 60-line component results in ~350 lines of generated boilerplate code), it's become hard to maintain and extend. While I'll still be using source generators, since it's a great feature, I'll try to tone it down and only generate the necessary "glue", keeping most of the framework logic in the project.

 

[Twiglard]

Hey.

It's been a while. Fortunately, now there's at least some progress.

Initial version of the pathfinding code now works.

---

Here's CMake code that splits off DWARF debug info for llvm and gcc while preserving debugging capability, including on Windows. As you can see it requires a binutils-compatible linker but works with llvm-objcopy as well.

Edit: on clang+lld you can link LTO+PGO with debug info. Didn't work with binutils last time I checked.

You could improve it to work on .dll's/shared objects as well. I statically link dependencies with LTO so the extra complexity isn't necessary.

cmake code

set(CMAKE_RUNTIME_OUTPUT_DIRECTORY "${CMAKE_CURRENT_BINARY_DIR}/$<CONFIG>/bin")

# self = target name, assumed to be an executable
function(foobar_add_install_executable self)
    if(NOT MSVC AND NOT APPLE)
        if(CMAKE_RUNTIME_OUTPUT_DIRECTORY STREQUAL "")
            message(FATAL_ERROR "")
        endif()
        if(NOT CMAKE_OBJCOPY)
            message(FATAL_ERROR "objcopy missing")
        endif()
        set(exe "${self}${CMAKE_EXECUTABLE_SUFFIX}")
        add_custom_command(TARGET ${self} POST_BUILD
                           COMMAND "${CMAKE_OBJCOPY}" --only-keep-debug "${exe}" "${exe}.debug"
                           COMMAND "${CMAKE_OBJCOPY}" --strip-all "${exe}" "${exe}.nodebug"
                           COMMAND "${CMAKE_OBJCOPY}" --add-gnu-debuglink=${exe}.debug "${exe}.nodebug"
                           # BYPRODUCTS "${exe}.debug" "${exe}.nodebug"
                           WORKING_DIRECTORY "${CMAKE_RUNTIME_OUTPUT_DIRECTORY}"
        )
        install(FILES "${CMAKE_RUNTIME_OUTPUT_DIRECTORY}/${exe}.debug" DESTINATION bin)
        install(PROGRAMS "${CMAKE_RUNTIME_OUTPUT_DIRECTORY}/${exe}.nodebug" RENAME "${exe}" DESTINATION bin)
    else()
        install(PROGRAMS "${CMAKE_RUNTIME_OUTPUT_DIRECTORY}/${exe}" DESTINATION bin)
        if(MSVC)
            install(FILES $<TARGET_PDB_FILE:${self}> DESTINATION bin)
        endif()
    endif()
endfunction()

 

[Twiglard]

How are you doing wall and floor textures in tiled isometric games? I'm trying to think of a complete model of what a wall consists of. If you have a different or an already-proven wall model, consider discussing it or sharing reading materials.

1. Outer texture, i.e. wall on the outside of the building.
2. Wall's "depth", i.e. the part visible from above unless the wall is multi-storied (i.e. multiple walls piled on top of one another). Based on the outer texture.
3. Inner texture, i.e. wall's natural appearance on the inside of the building.
4. Room corner decoration, e.g. a vertical indentation or darkening, or other strip of color.
5. Outside wall corner decoration.
6. Window hole decoration???
7. Inner texture such as paint or wallpaper on a wall.

So when doing the inner wall (3), I'm thinking of a simple seamless texture that can continue from the West wall onto the North wall. Since the South wall uses the outer wall (1), there's no need to worry about a seam there.

As for corner seams (4), I'm thinking of a (semi-translucent, additively blended) scenery-like image that can be superimposed onto any arbitrary wall tile (because the tileset maker doesn't know how long the wall is going to be), with half of the image on the West tile and half on the North tile. Such that each of these halves is the size of either a full regular tile or half of the regular tile (I think Fallout used half regular wall tile size).

It's possible that wall's depth (2) requires a special variant for the corners so this is even more of a headache.

Possibly (1) and (3) could be unified into a single texture (I think Underrail does this?).

For (6), there shouldn't be a need to worry about seams toward the (1), (2) or (3) textures.

For (7), it needs its own inner corner decoration.

Fallout seems to be doing it like this at least. It has long textures that don't quite repeat, and can't be permuted across the length of the wall.

Then there are wall tilesets such as concrete blocks where tiles can be arbitrarily permuted across the wall's full width due to them having indentation on each tile's edges.

 

[The_Sloth_Sleeps]

How are you doing wall and floor textures in tiled isometric games? I'm trying to think of a complete model of what a wall consists of. If you have a different or an already-proven wall model, consider discussing it or sharing reading materials.

1. Outer texture, i.e. wall on the outside of the building.
2. Wall's "depth", i.e. the part visible from above unless the wall is multi-storied (i.e. multiple walls piled on top of one another). Based on the outer texture.
3. Inner texture, i.e. wall's natural appearance on the inside of the building.
4. Room corner decoration, e.g. a vertical indentation or darkening, or other strip of color.
5. Outside wall corner decoration.
6. Window hole decoration???
7. Inner texture such as paint or wallpaper on a wall.

View attachment 42434

View attachment 42435

View attachment 42433

So when doing the inner wall (3), I'm thinking of a simple seamless texture that can continue from the West wall onto the North wall. Since the South wall uses the outer wall (1), there's no need to worry about a seam there.

As for corner seams (4), I'm thinking of a (semi-translucent, additively blended) scenery-like image that can be superimposed onto any arbitrary wall tile (because the tileset maker doesn't know how long the wall is going to be), with half of the image on the West tile and half on the North tile. Such that each of these halves is the size of either a full regular tile or half of the regular tile (I think Fallout used half regular wall tile size).

It's possible that wall's depth (2) requires a special variant for the corners so this is even more of a headache.

Possibly (1) and (3) could be unified into a single texture (I think Underrail does this?).

For (6), there shouldn't be a need to worry about seams toward the (1), (2) or (3) textures.

For (7), it needs its own inner corner decoration.

Fallout seems to be doing it like this at least. It has long textures that don't quite repeat, and can't be permuted across the length of the wall.

Then there are wall tilesets such as concrete blocks where tiles can be arbitrarily permuted across the wall's full width due to them having indentation on each tile's edges.

One solution is just producing duplicate pieces and corner pieces. E.g. https://kenney-assets.itch.io/isometric-prototypes-tiles

Kenny has a solution, which I used for a prototype, textured my own over the top and made some experiments in Tiled.

I assume that much like tiled, you have the various tile layers and then freely placed object layers on top of that.

 

[Twiglard]

I assume that much like tiled, you have the various tile layers

Are you assigning layers to individual graphics pieces at tileset declaration time or each time the tile is placed?

 

[Bad Sector]

Added a tag system to Little Immersive Engine. Tags are part of "tagsets" which can have an optional context so that it can be easy to share tags between objects of the same context (in the video above all tags are in the "World" context). Assets can also be tagged (in the "Assets" context) and it will be possible to filter assets using their tags, though at the moment i still haven't worked on the UI for that.

The idea behind the context system, aside from avoiding unrelated tags from showing up in the editor, is that the context keeps track of all tagsets and their owners so it will be fast to do searches like "give me all objects with the Foo tag" without having to actually go through all objects (99.9% of which wont have any tags).

 

[The_Sloth_Sleeps]

I assume that much like tiled, you have the various tile layers

Are you assigning layers to individual graphics pieces at tileset declaration time or each time the tile is placed?

Everything design time for static objects. When you are designing the level the tile is assigned to a layer. All objects within the layer are Z sorted.
However any layer on top of a layer means ALL objects in that layer no matter what are drawn on top of all object in the under layer.

 

[Bad Sector]

More tags, making it work with multiple selected objects and with undo/redo working as expected turned out a bit more complex than originally thought (also came against a bunch of assumptions on how property editing works that i made ~3 years ago when i wrote some of that code - not only had to refactor a bunch of stuff but also ended up implementing and throwing away a couple of alternative undo/redo approaches for tagsets until i made one that works).

Also added a dialog box for selecting objects by their tags.

Amusingly the whole thing started because i wanted to add tags to assets, but aside from adding the tagset property itself to the asset files, i haven't yet worked on that :-P.

 

[Bad Sector]

More non-visually-exciting stuff :-P.

I added the asset tags mentioned previously (and the asset tags are saved as part of the asset file but without needing to deserialize the asset itself, they're treated as metadata discovered at startup - this is so that the gameplay code can enumerate assets using tags to find, e.g., weapons, items or other assets that in theory could be extended by mods without needing some central table or something like that).

Another thing i just worked on - and shown in the screenshot above - is the ability to open multiple asset browsers in the main form's sidebar. The idea behind this is to have 2-3 browsers for different things, like showing all the materials in a flat view (no thumbnails yet, i plan on implementing this at a later time) or having two separate folders.

The non-visually-exciting stuff will still continue though as there are a bunch of other things i want to do in the editor before touching the engine side again.

 

[Bad Sector]

This one took me the entire day and is a change that affected several parts of the engine: added asset thumbnails in asset browsers.

There are different thumbnail generators per asset class (texture, mesh, etc), some use the asset directly to create an image (e.g. texture thumbnails are just scaled down texture images) and others do a render to an offscreen buffer first (mesh and material assets render a temporary world, world assets render the world directly). For the renders, asset editors can setup the camera position and direction (though a default is also available using the render scene's bounding box). The images are cached as PNG files next to the asset file so that the assets do not need to be loaded for their thumbnail to show up in the browser.

 

[zwanzig_zwoelf]

Semi-actively working on Demon Lord Reincarnation II. Took a bit of time to analyze the criticism of the original game and what I wanted to do but didn't due to the need to test the waters first and I think I've found the right spot.

So far I've added an equipment system that works similarly to Makai Toushi SaGa, meaning that you have a limited number of equipment slots (4), but it's up to you to decide what to do with them.

1 weapon + shield + armor + accessory? Ok.
2 weapons + shield + armor? Sure, an alternative weapon won't hurt.
4 weapons and no armor to become a jack of all trades? Sure, go for it.

Right now I'm designing passive abilities attached to each class to make party creation and dungeoneering more interesting. Stuff like cartography, max chance to unlock treasure chests without triggering traps, ability to detect hidden doors, etc.

The skill system has also been revamped to allow players to forget skills they don't need at this point or seal them forever if they don't fit their playstyle.

Stat growth and damage formulas have also been reworked from scratch since progression didn't feel quite right in the original, making players feel they weren't progressing. That, and I reworked encounter scaling into a more subtle system that only increases the number of enemies over time, while enemies have a large, yet static set of skills instead of a limited number of slots filled from a large pool.

I'm confident it'll be a massive improvement over the original and a much better game on its own.

 

[what am i doing]

This one took me the entire day and is a change that affected several parts of the engine: added asset thumbnails in asset browsers.

That took the Gnome desktop environment team 18 years to implement.

EDIT: more UI work and refactoring and general project cleaning. Replacing a lot of the direct path usage with function calls that take a file name and prepend the appropriate path, so that if a path changes, it only has to be changed in one location. This came up because I want to start moving some of the folders around and realized it would be better to take care of this when converting to call-based paths was still relatively painless.

 

[Bad Sector]

That took the Gnome desktop environment team 18 years to implement.

Assuming you mean the file picker AFAIK it wasn't a technical issue, someone implemented a better one more than a decade ago :-P.

 

[beardalaxy]

Well, I can't really SHOW you anything, but just wanted to say that I have finally finished making every side quest for my game, God's Disdain. 50 side quests in total, that appear all around the map, at different times of day, and even some that change depending on how long you've been playing the game or what decisions you've made throughout it.

I actually had a cool moment where I decided to walk outside after testing the tavern system and just see what happened. I went into a shop I didn't remember making, and got a side quest from the guy that was inside. I discovered something in my own video game within only a couple minutes of playing. Pretty wild. Not saying the whole game will be like that, but I imagine that I've done a good job in giving players at least somewhat interesting things to look at and seek out often enough.

Now, I'm onto making the non-quest-related NPCs. You know the ones. They give you info about themselves, about the town you're in, other stuff related to the lore, some advice, and even react to your adventures and how they have been affected by them. I have 99 of these little guys to make, and I just made 4 today, so I reckon it'll only take me a few months before they're done and we're one step closer to beta! My goal is to finish 5 NPCs a week, and to have the beta ready by the end of 2024. If I can release this game before the 10 year development mark, which would be October 5th 2025, that would be just peachy.

As far as the split between NPCs who have an involvement in quests and those who don't, I'd say it's about half and half, although the ones who don't have direct involvement may still be affected by them in smaller ways, like a change in dialogue.

 

[Zanthia]

I finally got to the point where I need to develop my Inn system in order to finish up a quest and an NPC. There are two ways I could go about doing it, and I wanted to ask you guys for your opinion.

What did you decide in the end? I would have voted for a hybrid: ask for your room at the bar and wake up by your bed.

 

[beardalaxy]

I finally got to the point where I need to develop my Inn system in order to finish up a quest and an NPC. There are two ways I could go about doing it, and I wanted to ask you guys for your opinion.

What did you decide in the end? I would have voted for a hybrid: ask for your room at the bar and wake up by your bed.

That's exactly what I did lol. I asked in a few other places and someone else suggested that idea!

 

[Nathaniel3W]

Tons of updates for Septaroad Voyager. I'm going to try putting together a big update once a month from now on.

I have a separate thread for Septaroad Voyager. I'm not sure if anyone actually likes seeing updates there. I'll spam my stuff there just in case. But I'll try to focus more on development stuff here.

Portraits and locations made with some help from AI. I start with a piece of art I kind of want to copy, put that through img2img, and then do lots of inpainting with lots of variations, then I go with the best version, and iterate on that. Here's what the new character selection screen looks like with the new art.

Working on a new map in Inkarnate. I'm populating the areas of interest along the coast of the inland sea first. I plan on letting the player take on some quests to clear pirates or sea monsters or whatever to allow sailing a ship to get from one port town to another.

New building styles. I'm basically remixing the same shapes and textures over and over again. But here the roof angles are sharper, with exposed beams at the ends of the roof, and I'm using more of the wood texture on the walls than I do elsewhere.

For reference, here are the same basic shapes and same texture atlas being used to create buildings in a different town:

Extended camera functionality. Here the level has a built-in boundary for the camera. The camera won't cross the boundary. And if your character reaches the 20% of the screen on any side, the camera zooms out automatically.

Guard NPCs. Getting this started wasn't so bad, but there are a lot of little edge cases and complications that I kind of expected I would run into. Nothing is ever really easy.

 

[Russia is over. The end.]

Any advice from people who manage to put in consistent work into their project for years (5+) without burning out, losing interest, despairing ? Looking at you, Nathaniel3W among others.

 

[Nathaniel3W]

Any advice from people who manage to put in consistent work into their project for years (5+) without burning out, losing interest, despairing ? Looking at you, Nathaniel3W among others.

So the thing that kept me going at first was the feeling of absolutely despising my day job. I felt completely stuck and hopeless. Even when things started turning around for me, and I was getting offered the assignments I wanted, I still felt like I was going to be miserable. I felt like the only thing that I was actually making progress in was my game; every other aspect of my life was just treading water. I think it was that feeling, that game dev was the only thing that I even could make progress in, that made me stick with it at first.

Eventually game dev became my job. This is what I do. It's not especially fun or rewarding or interesting. The stuff I'm doing isn't cutting edge. It's pretty slow and tedious. But I keep doing it because it's my job. A hobby is something you do when you feel like it. A job is something you do whether you want to or not because it pays the bills.

That's probably not very inspiring, but that's the absolute truth from me.

 

[The_Sloth_Sleeps]

Any advice from people who manage to put in consistent work into their project for years (5+) without burning out, losing interest, despairing ? Looking at you, Nathaniel3W among others.

So the thing that kept me going at first was the feeling of absolutely despising my day job. I felt completely stuck and hopeless. Even when things started turning around for me, and I was getting offered the assignments I wanted, I still felt like I was going to be miserable. I felt like the only thing that I was actually making progress in was my game; every other aspect of my life was just treading water. I think it was that feeling, that game dev was the only thing that I even could make progress in, that made me stick with it at first.

Eventually game dev became my job. This is what I do. It's not especially fun or rewarding or interesting. The stuff I'm doing isn't cutting edge. It's pretty slow and tedious. But I keep doing it because it's my job. A hobby is something you do when you feel like it. A job is something you do whether you want to or not because it pays the bills.

That's probably not very inspiring, but that's the absolute truth from me.

I have hardly been consistent so I cannot give advice on that, but when I take breaks it not really burn out but more that I cannot manage work and game work at the same time. But I keep coming back.

I really strongly agree with the hate aspect giving you strength. Hate has also been a strong form of inspiration for me in many walks of life and also game dev.

 

[Russia is over. The end.]

I consider myself a humongous failure at going solo dev.

I've designed most of the game, with over 30 pages of formulas and concrete game design info. I'd say 90% of the game design/RPG system is in, at least the foundational, hard stuff is in. Adding mechanics on top is easy now.
I've spent 2 years coding all the tools, so I have a world editor, which is pretty complicated and does a lot of fancy shit, an item editor, a mob editor, and a bunch of other editors. They're not just editor extensions in Unity, they're standalone and allow for non-technical users to work in them, without ever installing any shit at all. Other than the quest editor, I'm done with the editors.

I have a custom engine, by the way, which was certainly a big mistake.

I have coded the movement, inventory, equipment, communication, battle logic, ability logic, buffs, and some other minor stuff for the server. It's all replicated to the client. The client displays it all.

But I'm not even half way there. I didn't fill the world with anything. There's a shitload of mechanics still necessary to be coded in. And I'm tired, I can't look at it anymore. Whatever enthusiasm I felt, and it was a great one, has evaporated a long time ago. I look at it, and feel nothing but exhaustion. I haven't touched the code in almost a year now. I can't bring myself to.

And at the same time, I know time is running out. AI will start doing all the shit that I did, and will render all my efforts useless.

I have enthusiasm for other projects, but I know if I start them, I'll burn out years before release and then it'll be the same thing. Also if I start a new project, this'll be really it for the old one. And I can't bring myself to say it's over. So I'm paralyzed into doing nothing.

I feel like I'm a big fucking pussy. Men should have more control over their actions.

 

[Zanzoken]

If you haven't worked on your project in a year and you're still feeling sick at the sight of it, then just call it dead and move on. Sucks to admit you failed, but it's better than pointlessly spinning your wheels.

With that out of the way, take the tools and knowledge you have and come up with something new that you can actually complete in a reasonable amount of time. Don't try to make the biggest fucking thing ever... figure out the core gameplay loop, write down all the features you want, and then cut everything except what is absolutely necessary for the concept to work. Then you're ready to start.

 

[Kev Inkline]

Only those people won't fail who never attempt anything. It's a platitude yeah, but it's also true.

 

[what am i doing]

I consider myself a humongous failure at going solo dev.

I've designed most of the game, with over 30 pages of formulas and concrete game design info. I'd say 90% of the game design/RPG system is in, at least the foundational, hard stuff is in. Adding mechanics on top is easy now.
I've spent 2 years coding all the tools, so I have a world editor, which is pretty complicated and does a lot of fancy shit, an item editor, a mob editor, and a bunch of other editors. They're not just editor extensions in Unity, they're standalone and allow for non-technical users to work in them, without ever installing any shit at all. Other than the quest editor, I'm done with the editors.

I have a custom engine, by the way, which was certainly a big mistake.

I have coded the movement, inventory, equipment, communication, battle logic, ability logic, buffs, and some other minor stuff for the server. It's all replicated to the client. The client displays it all.

But I'm not even half way there. I didn't fill the world with anything. There's a shitload of mechanics still necessary to be coded in. And I'm tired, I can't look at it anymore. Whatever enthusiasm I felt, and it was a great one, has evaporated a long time ago. I look at it, and feel nothing but exhaustion. I haven't touched the code in almost a year now. I can't bring myself to.

And at the same time, I know time is running out. AI will start doing all the shit that I did, and will render all my efforts useless.

I have enthusiasm for other projects, but I know if I start them, I'll burn out years before release and then it'll be the same thing. Also if I start a new project, this'll be really it for the old one. And I can't bring myself to say it's over. So I'm paralyzed into doing nothing.

I feel like I'm a big fucking pussy. Men should have more control over their actions.

Take a break, maybe do a very small project (something with a specifically constrained scope and time allocation so it stays under control) then come back to this.

You shouldn't consider yourself a failure at going solo dev because your project took a long time. That's the nature of going solo. It's a ton of work and it takes a really long time and if you're making something non-trivial it'll probably be many years. (I am now a little bit past three years into making a spiritual successor to Victoria 2 and it is nowhere near playable, and in that time two open-source V2 projects have shown up plus the series I thought was dead got an official sequel out of nowhere - so whatever earning potential it had is basically gone. I am still continuing.) If nothing else you should be happy that you have made progress in realizing your goals - even if it's not done yet and won't be for a long time - instead of just dreaming.

I do know exactly how you feel with regards to not touching code and not wanting to. Personally, I've found the best way to get myself back into programming whenever I've shifted into another area (to avoid burnout) is assign myself a very specific task for the day and work on it. Just that task. Yesterday I assigned myself the task of fixing tile importing to have a modal popup asking for confirmation, accessed by a menu bar, instead of just one button sitting in a placeholder window that instantly did the task without confirmation. Today I finished that. It's a small task, and took longer than I would have liked due to needing to figure out some aspects of how Dear ImGui works, but I have a ton of things I need to do and if I try to tackle them all at once I just won't feel like touching it at all, and small progress is better than no progress. One thing at a time. Identify something that needs doing, make that your target, don't worry about anything else. Once that thing is done, pick another thing.

AI will not do the stuff you've implemented, especially not programming. It's very useful for some things (in my opinion, mostly audio-related) - it's not and probably never will be a button for braindead retards to make a good game with.

EDIT: also, your mileage may vary on this, but personally I often find it a lot easier to sketch out my program logic with pen and paper, and only start writing the actual C++ code once I've determined how it should work. Separating the mental work of building up program logic from the actual process of typing makes it easier for me, especially when I'm in a mood where I don't really want to be staring at my IDE.