The intention is to incorporate the dungeons into official releases. This will attract players, and perhaps some developers. This means you can create full-length dungeons and be confident that they will continue to appear in continuing development.

I would say, the most direct/guru way would be to fork the git repository and you can work on your own dungeon in your own fork, commit, push changes to your remote, and then pull request the changes into attnam/master. However, as I write this I can see that with the status quo there might be long times between releases. I feel that there will be more frequent releases as dungeons reach the cut-and-polish stage.

You will want players to play test your dungeons as soon as you make changes. The players don't use github, that much is obvious. But I would recommend you use github as the vehicle of ultimate authorship, because it will enable you to show the work was yours, and help you track changes etc. It also means I can see it if you get stuck, or if there is a bug. When you feel the code is ready you can make a pull request, I can merge this into the main line and do a release. In this way it will be released to the general public.

For all github's technical usefulness, it will not get you the rapid feedback you require. The forum here is a good development nucleus with plenty of the game's legendary power players who drop in on us from time to time. When I did CLIVAN I found lots of people coming out of the woodwork to push my dungeons to their limits and the feedback was awesome! They smashed through everything and found all the loopholes, which was embarrassing, but humbling that people took the time out to review my work.

I expect if you get a thread going on the forum here you will get some eager followers. As you develop, attach your dungeon data files here and explain where they go so people can test it out. If you don't want to use git (the learning curve can seem steep, I'll admit) and you start out in this way, then that is fine too. You can even jump into github right at the end and push all the changes in one go. I know you already have a fork though...
The first part will be to flesh the thing out using your imagination, and writing the dungeon script (or copy-pasting a lot and changing things). It will be a lot of solitary hours, keep a notebook handy, graph paper is good too.

I know right? I'm working on a guide on the wiki right now

Well, after some work, here it is: https://github.com/Attnam/ivan/pull/130

I finally figured out how to populate the worldmap in the right way. You start with the main continent and work through the available locations you sampled. Then you join up the terrain type of the available location with the dungeon you need to place, by iterating through the entire list of dungeons. Then you eliminate those locations which are occupied, and those dungeons which have been placed. Then move on to the next nearest continent until you run out of continents.

Below is a screenie showing the diversity of terrains on which you can place the locations (they all look like New Attnam in this example).

Ok, so what does this all mean?

This means in IVAN 0.50.6, you will be able to add your own dungeons on the worldmap, using the script files alone!!!

Here it is

I've had that message once I wonder what it refers to?

I have been doing more; this weekend I built some machinery for collecting data into vectors of structs (not sure if this is efficient). I transplanted some code normally used for the prototype system, and it rather unexpectedly worked "right out of the box". Not exactly "flow", but a pleasing experience nonetheless.
I very nearly have something ready to deploy to the main line of development. I think it will be another weekend before I can do this. My current problem is to find the best way of filling the available spaces with locations according to a combination of terrain type, proximity to Attnam and what continent. I figure I will find a way of filling up all the available spaces on the main Attnam continent first. Then once it is full, then move to the next nearest available spaces, even if they are on another continent. It's a real pain because you can get underflow and overflow if the number of terrain types differ greatly between available types and the types to be placed on the world map. I may just have to be expedient and make some stern simplifications. The next adventure is how to use c++'s Erase-remove idiom to remove those vector elements of locations that have already been placed.

After all this I'll need to make a how-to guide for adding new dungeons without compiling.

I am toying with the idea of adding a new swamp terrain type called "morass". Does anyone want to draw a picture of it?
For that matter, would anyone object to new drawings of the world terrains? There's no deadline... Azba? blob?

Yep, the rain works best when you lay the burning items on the floor, individually and not stacked. I don't think the rain ever makes it to your inventory. Not sure how this can be improved.
I know about not being able to dip things with fluid already on them. I don't know how best to tackle that. Have you tried vomiting on it?

I love that this win is in the dev version! Good thinking using the acid explosion, it must've taken some patience and careful planning. Did you get hassled by the fire system? Frightened by ghosts?
Did you find the application stable enough or were there moments when you feared the game might break at any minute?

This weekend I was busy porting the poisson disk sampling algorithm from a python model to c++. I learnt that the std::vector type is a good proxy for a python list, because you can use push_back() and pop_back() to add or remove things from the vector in much the same way as you would use append() in python, and much more besides. I was also reminded that the ^ operator is the logical XOR in c++, and not raising to the power (stupid matlabthink).

We're getting closer to making it possible to contribute dungeons, just check out the attached file! I committed the code for posterity, but it is quite scrappy so I will tidy it up as I come up with a way to integrate the sampler with actual user-defined dungeons (the easy part).

I wonder if the monster generator has increased the attributes of the golem after generation? The generator can do this to player ghosts to increase their base attributes relative to the deceased.