Why not terrain that interacts?

But who ever build dox instead of tanks because they were next to a forest? Forests could be more of a natural barrier early game requiring them to be reclaimed or burnt down and reclaimed which would be quicker. If you had swamps and rocky areas and other things affecting movement speed then you'd have a game which is a lot more tactically rich than what we have now. It would be cool to have to burn down a forest and send in engineers to reclaim the mess so that you could get your tanks through.

I for one always assumed that forests would actually block pathfinding instead of just making units scatter around a bit. The still move through trees as if they were not there, as they are actually passable.

It used to block unit movement if I remember correctly, but forests are so dense that it really really sucked to spawn inside of one.

Remember whenever they added being able to set fire to forests- I swear tanks couldn't get through but bots sometimes were able to shimmy about within. (Could be wrong, maybe just memory playing tricks. ;P )

I think this is simply because of the large technical challenge involved. You either put it on your wishlist, or you throw money at it, and i don't think Uber are can throw much more money at PA - they're still developing it as it is, so for a feature like this which isn't /that/ great imho, it's just not going to happen.

Last I read about it, the pathfinding is done using a voxel map around the planet, which is quite an interesting solution but also offers a few technical challenges. Complicating it by having to make it interpret ravines as pathable for some units instead of just simply unpathable, is a tall order. I think

Actually not bad at all. That would only require an invisible bridge above the ravine, which is only traversable for units having a certain attribute.

Getting it to look right would be the difficult part, respectively preventing units from jut stopping dead midair on the "bridge". Pathing wouldn't have any problems with this.

Remember, the navigation mesh, which is the base for the voxel structure, can actually contain quite complex topologies.

It's this.

I really think you're underestimating how much time this would take developers to implement, and how much time they have available. They must already have a roadmap of what they want to do, we know they're working on a big update to GW and other things that they've mentioned in a recent developer's post. I'm not saying it is impossible, I'm just saying it is quite difficult to do, arguably there isn't much reward for doing it, it isn't that important in the grand scheme of things (the competitive community I'm sure doesn't care much about this feature)... etc - there are other things that are more important to do for the PA, imho.

It's all about money and a large enough RTS player base.
Have as many players as Call Of duty like games and RTS genre will become a gold mine for editors.
Unfortunately it's unlikely to happen.
Developping such a complex RTS engine takes a huge amount of time and requires much more tech skills than developping on top of an allready existing graphic/physic engine. Thus it requires lots of money.
Stardock has been inspired and lucky enough with galactic civilization not to get prisoner of a very big editor and to make lots of money to stay independant and develop the games they want to develop. I may not have all the insights, but it probably explains why they were able to bring "Ashes of Singularity" on to the table.
I'm nearly sure that Uber devs would possibly make lots of things differently if they had the required money. But in the end it's all about business. I simply hope they have other projects to pump money into the studio, because they are very talented and the game industry needs them.

@doud Space Engineers recently gave a good example how else it can be done.

The point is:
With the few devs left at Uber, dedicated to PA, you can't expect them to develop as fast as the user base would actually expect it. PA was also never expected to be "that one" ground breaking RTS of its decade, but it still has a quite solid architecture which would allow for much more.

Currently, the biggest drawback is, that the promises on mod support haven't been kept. I'm not talking about trivial stuff such as modding the 2D components of the clients UI, or basic unit conversion mods, but actual full conversion mods where you can make use of all the engine has to offer internally.

I already hinted it in another thread, eventually PA will follow the same fate as the original TA. Nice ideas, but long term it's not going to be the closed source implementation the userbase is going to use.

they released their game source code to public by yesterday ! Impressive .... If only Uber could do the same ...

Yes, they released the source. And people are still paying more for Space Engineers than they do for PA. That game has a giant modding community and even fully fleshed server side scripting capabilities. (Now no longer just scripting, but actually fully custom server software...)

The reason why they did is quite simple: It gives the core developers the option to focus on features THEY want, with the option to port user developed features back into the core game.

And I'm not just talking minor fixes to the GUI, but features from actual full conversion mods.

In the same way as human resources would have been a full conversion mod for PA, with the benefit of porting features developed for HR back into PA.

You never know they might eventually release the source. Uber (especially John Mavor) aren't anti modders, I think you need to understand it's all relative. Many games companies go out of their way to prevent modding, PA is fairly high up on the scale of mod friendliness imo. Yes there is *lots more* that could (and hopefully will) be done, and I think if it gets to the point that Uber folds they'll probably release the engine to the community (I mean they did bid on TA with a view to do exactly that). In the mean time I guess they're just not comfortable releasing everything at this time.

There are a few significant gotchas with PA....

One of them goes by the name "Coherent". The other one goes by the name "fmod". Both have very restrictive licenses of their own, so PA can't go open source while it makes use of these.

Well, that's at least true for the client.

I haven't seen any such software in the server, so the source for that one could actually be released. And that will be necessary sooner or later.

I wonder how bad that actually is. I think I read that coherent is starting to open source some of their libraries, so they might not be that bad of a roadblock.
From a more technical perspective I would also think that it is possible to make the libs a "black box" that is not available as source, only as binary that can be compiled against. So basically they would not need to distribute more than they do now as well. But I dunno, not an expert on how to build native applications.

I doubt Uber has access to the source of the libraries either, it's s much a block box to them as it is to us. But in addition to the library itself, there is always a header. That header is usually subject to the NDA as well, as whoever has the header could just link the library against his own application. Apart from the header, there is only the documentation left as the only non-public component.

I guess they could make a wrapper library that hides the real API behind a custom one made for PA. So the wrapper is build against coherent and then the wrapper is only provided as binary with a header, but that header has nothing to do anymore with the original coherent library.

Though I doubt coherent is actually that "secret", there is public documentation: https://coherent-labs.com/Documentation/cpp/index.html
Coherent also isn't exactly opposing open source, as coherent for mobile is in fact open source: https://github.com/CoherentLabs/CoherentUIMobileOpenSource

I doubt piracy is a big issue for coherent, nobody in their right mind would pirate a framework like that and then make a commercial application based on it.

There certainly are ways around all license and legal issues, it is just a matter of time and money that needs to be spent on it.
Those two resources dictate a lot

And one of those legal issues is stupid software patents, which is which bgolus stated that open sourcing PA is a confirmed no...
https://forums.uberent.com/threads/source-code.64911/page-3#post-1014834

So that's basically "game over" for PA then. That leaves only one option, rebuild from scratch, open source to begin with, hosted outside the US.

Seriously, that must suck to be a software company in the US. Right now, I can't even think of any other country which would even have allowed trivial software patents.

Yeah but even that is just a question of work and money. Surely the stupid patent system in the US makes it unreasonably more expensive, but still a question of time and money nonetheless.

Well, the fact that those stupid software patents have a lifetime of 20 years (I think?) means they might get sued until 2033, right? I'm not sure if that is a risk they would want to take. So, it's not only time and money, but also a risk that Uber need not take.