Actually, it's caused by the combination of rails on a double-size grid and train cars having effective length 7. If cars were changed to be 5 or 7 long, or the inter-car area increased to 2, so effective length is even, this would resolve the problem. Meanwhile, building a double-headed train with ...

I'm seeing a lot of clash between a camp saying 'Bots are OP' and a camp saying 'Bots are harder than you think'. Are there any ways to change bots so their complexity becomes more obvious? The first aspect I am thinking of here is the current charging mechanic. This appears to be something only hea...

I could see having separate circuit conditions for picking up and dropping, though that could be too confusing to put in the main game.

A bridge which goes over anything would necessitate at least 2 positions on the Z axis, or anything standing below the bridge would collide with a train passing over it.

I think windows' builtin driver goes up to 5 buttons and a wheel. Mice with more buttons tend to have special drivers. Linux tends to go higher, limit probably 16 or 32?

To empower belts, I like the idea of introducing crates/pallets, which contain a full stack of any item, don't stack themselves, and cannot be carried by bots. Packing or unpacking could be done in a normal assembler. This potentially also makes belts more UPS-friendly, since the number of items mov...

Multiply the amount in the base by -1, put it on the red wire. Read the train contents on the green wire. Attach both of them to a filter inserter in set filter mode.

If you have a smartphone, it almost certainly includes a calculator app as well.

Artillery auto-targeting?

Did moddable map generation parameters make it into 0.16?

I could see giving some penalty to paths through a two-way signal.

Barrels only stack to 10, so stack inserters can only move 10 barrels per swing, which makes it 3.03s per full wagon or 33k/s. And it is the full wagon that is key here. 40 stacks of barrels means 40 swings to move everything out, plus 40 swings to move the empty barrels back in. If you store fulls...

I would appreciate even more different wagons for different types of cargo. We got box cars and tankers. And the box cars should keep beeing usable for any type of cargo (liquids after barreling) but should be the less efficient option when there is a more specialized wagon available. What about fl...

How about a new item ('loading funnel'), needing a belt and some iron, which inserters can insert onto to achieve compression. Unlike the suggested 3-to-1 compressor, this would only work with items inserted into the funnel. An inserter would insert into the funnel (holding up to 2 unstacked items, ...

I like the idea of linking to biome, or otherwise spawning more iron in some directions and more copper in others. By later game, you're training it around anyway, so if you need iron, expand in a direction which has iron, and if you need copper, expand toward copper.

Well, if you're offering bindings for C#, wouldn't that mean any other .net language would also work?

I may be wrong, but based on the FFF it appears that you have three different processes that update a single variable, but that variable is shared between each. (...) I also suspect that this isn't the correct path for parallelization any way, you'll likely find that each step itself could be paral...

creating/reaping threads costs time too. I do hope you're not doing that every tick?

Notwithstanding the interesting discussion about whether your analysis is right, I have a suggestion for achieving parallel thread update isolation at whatever granularity you need without moving data as entities move. Given N threads that will be updating in parallel some object type T, create N s...

Regarding skipping stations: How about a set of conditions on whether to visit a station, in addition to the wait conditions? Normally this would be an empty list, meaning the station is visited. Adding a rule like 'Fuel less than X MJ' or 'Cargo of Iron less than 8000' could make a station be skipp...