Construction chests

[Zappes]

It would be very useful to have chests that behave like a provider chest, but exclusively provide items to construction bots. When building outposts, one could place requesters for the required materials, feed their contents into those "construction chests" and then have construction bots take their building materials from those.

While it sounds tempting to include the behaviour of requester chests as well, I think that would take away too much of the challenge. Having to use a separate requester and construction chest with an inserter inbetween forces you to implement functionality like refill hysteresis using the circuit network, which preserves the challenge while still providing the basic idea of having a stockpile exclusively for construction.

[mrvn]

I think I never had any request chests for construction materials. You need inserters and belts for science packs, which you usually set up to be produced locally way before you have logistic robots. Can't think of any other construction materials you would use in production instead of construction.

[Zappes]

This is not something you'd find too useful in the "regular" game, it comes into play only if you do certain late-game things. In my current game, I built up my base until the first satellite was up in the air. After that I could not really keep on going with that base as it wasn't scalable enough for expansion to a true megafactory.

I then built a railway to the far east and started a new factory there, one that's designed specifically for feeding an array of rocket silos. This means that I don't mine or produce anything there that doesn't lead to this goal - especially no belts, splitters, inserters, furnaces or rails. I still produce those in my original base where I have the proper infrastructure in place. Practically everything I build in the new base get's blueprinted and built by construction robots, and that's where the idea of a construction chest comes into play as I don't want those robots to keep flying the really long distance between bases for every piece of rail or belt they want to put on the ground.

Sure, I could use a train to transport whatever I need to the new base and put everything in passive provider chests, but I don't really want to do that for several reasons. I'd very much prefer it if the logistics bots took care of that - but with the means available right now, there doesn't seem to be a sensible way of doing it.

[ssilk]

Added to viewtopic.php?f=80&t=18093 Roboport/Logistic Network/Robot enhancements

This is often suggested. Like here viewtopic.php?f=6&t=2680 Construction Chest
viewtopic.php?f=6&t=8877 Maintenance/ Construction Logistic Chest
Even by me:
viewtopic.php?f=6&t=2912 Chests for construction bots


I think I changed my opinion about this. I played several games like so, but I don't found any fun in this, cause it is so slow to let the robots bring the items.
And even with this chests: The more you request, the longer it takes to produce all items and fill all the construction chests with the requested stuff. And even then: You will still miss one item-type you forgot. And then you need to wait again.

Hmm.

After playing so many games I can say definitely, that it is much more fun to have another, much faster type of transport. Currently I'm testing the LTN ( viewtopic.php?f=97&t=36976 ) in depth, and that is a hot candidate for doing such things (still far away from being perfect, I miss for example a simple interactoin between Logistic Robot Network and Logistic Train Network). But it is really much more fun to request the items this way instead of "just put down a chest".

[Zappes]

Hmm, I somehow failed to find these other posts when I searched for this prior to posting my request. Sorry for that, but at least it's in the list now.

[bobingabout]

I don't think we need a specific construction chest, but it would be nice if Construction robots were able to take items out of a chest that is within the construction zone, but outside of the logistics zone.

[mrvn]

This is not something you'd find too useful in the "regular" game, it comes into play only if you do certain late-game things. In my current game, I built up my base until the first satellite was up in the air. After that I could not really keep on going with that base as it wasn't scalable enough for expansion to a true megafactory.

I then built a railway to the far east and started a new factory there, one that's designed specifically for feeding an array of rocket silos. This means that I don't mine or produce anything there that doesn't lead to this goal - especially no belts, splitters, inserters, furnaces or rails. I still produce those in my original base where I have the proper infrastructure in place. Practically everything I build in the new base get's blueprinted and built by construction robots, and that's where the idea of a construction chest comes into play as I don't want those robots to keep flying the really long distance between bases for every piece of rail or belt they want to put on the ground.

Sure, I could use a train to transport whatever I need to the new base and put everything in passive provider chests, but I don't really want to do that for several reasons. I'd very much prefer it if the logistics bots took care of that - but with the means available right now, there doesn't seem to be a sensible way of doing it.

So you don't want a construction chest. You want a chest that the logistic robots refill when construction robots take stuff out of it.

Simple: Place a passive provider chest, inserter, chest, inserter, request chest in a row and request the items in the request chest. Now for the hard part you need some combinators. You want to only move items into the provider chest if the request chest is full and only move items out of the request chest if the provider chest is empty (hence the extra chest in the middle as buffer). Otherwise logistic robots would endlessly move items from the provider chest to the request chest instead of fetching new ones from your first base.

Note: you can program the request chest by what items are missing with some combinators.

Note 2: This works much better if you don't connect the two bases by logistic networks. If you use trains, belts or simply an inserter to move items over a gap between the two logistic networks then you don't need complex combinators.

[Zappes]

@mrvn

Splitting the networks wouldn't really work for me, i'm afraid. The setup with the combinators and different chests would be an option, I guess - albeit a rather complicated one.

[TI-89]

Can you explain why splitting the networks wouldn't work? I have found it to be a pretty useful / flexible mechanic. At least, once I accepted that was what I needed to do. It certainly wasn't how I originally wanted to do things, and setting up chests can be a bit annoying. But other than that you can do a lot.

[ssilk]

Splitting the networks is a nice idea and I played a while with that, but it doesn't feel like it should be part of a game. It's just too awkward.

Have a look into those ideas:
viewtopic.php?f=80&t=18093 Roboport/Logistic Network/Robot enhancements
--> Logistic Network extensions

[TI-89]

It can be a bit awkward at first. Like I said, I originally had a much different idea of how things should work. And I've read most (ok, fine... half. But there's a lot to catch up on) of the threads on that page; many of them are the same or similar to thoughts I had while starting. In all honesty the extensions section has the least links that show being clicked, but I spent a lot of time figuring out how to distribute bots effectively. Again though I'm going to have to ask you to be more specific. I mean, for this mechanic to actually be removed from the game would require reworking logistics from the ground up. The ability to transfer resources from one zone to another using requester -> inserter -> provider ('logistics link' for brevity)* is just a consequence of how each of these works. So it seems to me (correct me if I'm wrong) that what you're saying is more along the lines of "I don't like it."

For example, let's look at one of the pictures from that thread:
Functionally, what changes if you replace each of the pink squares with a logistics link for the necessary resource? Absolutely nothing. And a logistics link places some restrictions on how things move between networks, whereas arbitrarily linked/layered (or colored) networks allow things to get pretty much anywhere automatically. But configuring something using a GUI to me seems far more against the spirit of Factorio than existing solutions. Plus this requires no changes to game implementation, whereas most suggested changes raise a lot of questions about how exactly bots would behave in the new system.

To me it seems a bigger problem is the fact that the construction network currently is basically a meaningless entity. It's just part of the logistics system. You can't provide items without a logistics network, you can't extend construction area without extending logistics area, and you can't meaningfully separate which ports are for construction and which are for logistics.

I don't think we need a specific construction chest, but it would be nice if Construction robots were able to take items out of a chest that is within the construction zone, but outside of the logistics zone.

This is essentially my thought (or at least a key part). I think wide area construction zone coverage should be much easier to achieve (cheaper port with large construction zone and small/no logistics zone would be nice). Construction network should link in a way similar to current logistics; bots could either grab from any logistics zone within the construction network, or have to be supplied as bob suggested. This also would make the research tree make more sense, because the construction network would actually be a usable mechanic outside of logistics. And explain the cost difference between construction and logistics bots -- the only real differences atm are stack size and ofc the ability to place entities -- because the logistics bots need the advanced circuits to receive a 'logistics signal,' whereas the logistics zone is invisible to construction bots.

In my usage, I've found that setting up a chest for each resource needed for everything you build is most of the annoyance in using 'logistic links.' If it worked automatically for construction, things would be a lot less tedious to build. And, as you noted:

I think I changed my opinion about this. I played several games like so, but I don't found any fun in this, cause it is so slow to let the robots bring the items.
And even with this chests: The more you request, the longer it takes to produce all items and fill all the construction chests with the requested stuff. And even then: You will still miss one item-type you forgot. And then you need to wait again.

While this solution is 'automatic,' it doesn't solve all problems for the player -- it actually suggests a problem for the player to solve (how to get supplies close enough to where you want to build that you don't have to wait forever for bots to place them). I haven't tried logistics train network yet, I've been trying to see how close to this I can get within vanilla. I'm not there yet, but I think it may be possible, if clumsy (see my response to viewtopic.php?f=18&t=41431&start=20). Maybe skip to page two and download the sandbox save I posted rather than trying to make sense of my rambling, it will likely save time. My thought is some combination of this 'logistics messaging' with same name stations. So then you could place train station named 'mining outpost' and send request, which would send a train that's pre-filled with supplies. Haven't quite worked it out yet though.
The other two changes that would make this more viable would be if construction bots could respond to requests if a chest is in construction zone only (no stack bonus -- transferring in any quantity would be infeasible, but could be a way to have inexpensive long range messages) and/or broadcasting of signals over logistics network (which would basically make the first option unnecessary).


*note that to the logistics system it doesn't matter how you move things between networks. You could also use requesters to load a belt or train that then goes to a different network.

[ssilk]

Instead of arguing I recommend you to play the "other styles" yourself. Maybe you can use also mods? After that phase I'm bet you changed your opinion. and if not we have a better base for arguing.

[TI-89]

Well what other styles are there? Currently I'm not aware of any great solutions to this. Logistics trains does seem like a good one, it's on my list to try (main reason I haven't yet is because there's still so much for me to learn in vanilla). But as I said once I accepted the way it worked I've found it's not actually too bad. If construction requests didn't have to be added manually I think it's actually pretty workable. Split networks are great for having a dedicated manufacturing/train station/supply cache/defense network, but it's extremely annoying when a section of concrete doesn't get laid because of the separated area. And I have no desire to add a provider chest to my manufacturing area just for concrete -- not to mention the fact that I don't have any construction bots in that network because it's dedicated to manufacturing.

I have a tendency to make posts longer than I initially intend to, my last reply was no exception. As I said I've spent a lot of time working with logistics, and I don't expect every player to spend as much time as I did figuring it out. So my intent was not simply to argue so much as ask -- what are the biggest barriers you have found to using this method?

EDIT - I'm not denying that I enjoy arguing, even more when it's about something I enjoy as much as Factorio . But I try not to argue without good reason. Here it's because I fail to understand why this: doesn't belong in the game (for labels of each area see diagram above).
BONUS: Just for fun, logistics bots are shared between all of these networks. I've removed all but one bot from each. I ask that you download this save and try it out. Place bots in one of the networks -- any of the networks -- and they will self distribute as necessary. If you try to scale up production, inserter speed will become a limitation (likely request size and number of bots will as well). The way bots/resources spread depend on how you arrange the links between networks, as well as the level of demand in each network.
Development of new mechanics requires balancing and then players will have complaints about these as well. This solution gives meaningful trade offs based on current mechanics.
And personally, the biggest obstacle in learning to use this effectively was the amount of time I spent thinking "there has to be a better way of doing this." One thing I forgot to mention in my first post: I think a good tutorial would work wonders, as well as provide a way to introduce usage of circuit network logic. There is certainly a learning curve, but what mechanic in Factorio does not have one?
If you try this out and still feel the same, I simply ask you explain to me why you feel that way. Because then I think I must be missing something which is very important.

p.s. I've never shared bots in exactly this way and I'm very happy with how well it worked. So I guess I got some good out of arguing even if you don't like it

[ssilk]

Well what other styles are there?

Difficult to explain. I recommend to inspire you by watching youtube or make multiplayer or use mods like bobs mod etc. The ideas are endless.

Here it's because I fail to understand why this: doesn't belong in the game (for labels of each area see diagram above).

Hm, I played a while with this and I think I can tell you what's wrong with this: It plays like if you have small isles of Factories on a very basic scale.

There are some things that goes objectively into the wrong direction:
- This goes too strong to pure micromanagement. And that should be avoided in Factorio. It's not fun.
- It goes strongly into direction of sub-factories (like the Factorissimo-mod). I'm not sure, if that is really a path that is "right". There needs to be some more experiments with the existing mods.
- This is too small to see, if it is useful in the long run. And there are some obvious problems when you make the "isles" bigger and bigger: There is a point, where this doesn't make any sense, cause you need to mix things with the size of your factory.
- And lastly: There are much better suggestions. I linked them above and namely I call here this one
viewtopic.php?f=6&t=8905 Overlapping Logistic Network II
It is in many points similar to this suggestion, but it's in "3D" and because of that it allows all the things, which I don't see here, plus the exchange between logistic networks is much simpler.
(I mention it not, cause it's from me, but cause it's objectively better in many points - of course far from being a perfect suggestion)

One thing I forgot to mention in my first post: I think a good tutorial would work wonders, as well as provide a way to introduce usage of circuit network logic. There is certainly a learning curve, but what mechanic in Factorio does not have one?
If you try this out and still feel the same, I simply ask you explain to me why you feel that way. Because then I think I must be missing something which is very important.

Well, Factorio aims to be this:

A wired, confusing, experimental, mind-blowing, always-something-todo puzzle game.

Compare that with your screenshoot.

[TI-89]

First off, I really appreciate the feedback. Thank you for taking the time to try it out and give me your opinion. Unfortunately I still feel the need to disagree on a few points.

I recommend to inspire you by watching youtube or make multiplayer or use mods like bobs mod etc. The ideas are endless.

I have tried Youtube videos, but I find it very difficult to sit through them, especially of others playing a game (I usually just end up opening factorio); if you have any particular videos you'd recommend I'd give them a try though. I have been meaning to try MP, so I can't disagree with that suggestion. But I actually used bob's mod to figure out most of these mechanics. Different size ports, faster bots, but nothing is fundamentally different as far as logistics networks.
I do like the smaller ports, though I'm not sure how far it should go. Part of the challenge is building non-overlapping networks with room to place chests between them. This forces trade offs in player decisions. For example, see how red circuits are passed through blue circuit network in the sandbox map, taking away from bots/circuits available for blue circuit production. But batteries I decided to place on a belt, making it invisible to the network. Belts also allow possibility of rotating shared goods between multiple networks (similar to this concept: https://www.reddit.com/r/factorio/comme ... p_modules/)

It plays like if you have small isles of Factories on a very basic scale.

This feels more like a problem with this particular layout than the mechanics involved. I mean, I just plopped down the bare minimum necessary to build each component. So... that's exactly what we have. Personally I wouldn't ever use networks quite in this way, for example there's no reason to have a dedicated one for iron/copper/steel -- I just thought replicating that diagram made a decent demonstration of how networks can interact.

There are some things that goes objectively into the wrong direction:
- This goes too strong to pure micromanagement. And that should be avoided in Factorio. It's not fun.
- It goes strongly into direction of sub-factories (like the Factorissimo-mod). I'm not sure, if that is really a path that is "right". There needs to be some more experiments with the existing mods.
- This is too small to see, if it is useful in the long run. And there are some obvious problems when you make the "isles" bigger and bigger: There is a point, where this doesn't make any sense, cause you need to mix things with the size of your factory.
- And lastly: There are much better suggestions. I linked them above and namely I call here this one
viewtopic.php?f=6&t=8905 Overlapping Logistic Network II
It is in many points similar to this suggestion, but it's in "3D" and because of that it allows all the things, which I don't see here, plus the exchange between logistic networks is much simpler.
(I mention it not, cause it's from me, but cause it's objectively better in many points - of course far from being a perfect suggestion)

I'm not sure all of these are 'objective' . But let's see:

• Personally I think this requires far less micromanagement than having to configure and name global networks, view layers, and assign chests to them. While setting up chests for lots of different resources isn't my idea of a good time, I don't think it's any more tedious than laying belts for resources or constructing train stops for separate resources. And there's a simple solution: keep the number of items which has to be transferred between networks minimal. With my method, setting up a new network doesn't require naming or assigning entities. You can blueprint the 'bot exchanger,' blueprint the roboport, even blueprint 'item sets' for things like a defense network that always need the same. Then drop in a single logistics bot and walk away to let everything else take care of itself.
• While it does allow you to subdivide areas of your factory into separate networks, I don't think it's very comparable to factorissimo, which actually lets you create more space within your factory. You make it sound like this 'changes the game' in some way, leading to the 'wrong path,' but this is just usage of game mechanics that already exist. I don't really understand this point.
• Yes, this version is small. But it was intended as a sort of 'toy problem' for the purpose of demonstration. Making it large makes it harder to understand. Things in a real factory should absolutely vary in size, but there should be real benefits and disadvantages in how you split things into large and small sections. I'm not totally sure where the consensus ended up with overlapping logistics (can bots move between separated networks or not?) but it seems like it gets rid of some of these trade offs for the sake of convenience.
• The exchange between networks may be simpler, but I think it moves towards being too simple. As I said I found myself on the other side of this starting out, named networks were one of my initial solutions (though I was leaning more towards 'subnets' than arbitrary linkage). But I think it moves towards being too simple. Should the game make all solutions simple, or should it provide you with problems to solve in creative ways? And personally I cannot come up with a single thing that this allows that cannot be accomplished with my method, other than bypassing networks between those you wish to link. Which is one of the trade offs I mention in limiting logistic functionality. I feel game components should be kept simple -- it works well with the theme of 'stranded space man,' and allows complexity to emerge according to the decisions made by the player. Infinite configurability may be convenient, but I don't believe it adds power to what can be accomplished, just ease in getting there.
  And I love the way the current logistics system is analogous to the internet -- each individual transmission is totally anonymous (packet/robot) and the system has no knowledge of the payload (data/resource) while it is in transit. And careful subdivision of networks using the method I have shown leads to a 'network of networks' (https://en.wikipedia.org/wiki/Internet -- read the second sentence)

Also you seemed to like the idea of sharing information between disconnected networks, but I have found hiding info and choosing what to show using circuit logic to be very useful. This is the essence of the train request system I linked before -- every resource shown in the large lognet is representative of a trainful ready to be sent (in theory -- as I detailed in the thread it's not perfect in implementation). Essentially as you said, things should vary in size -- small networks should be used for production/defense/train stations (movement of goods) and large networks should be used for long distance transfer of information (but this is not a hard rule, and is left to the player to decide).

Finally, this allows precise control over what quantities goods are transferred in and the rate at which they are transferred, through use of circuit logic and inserters. It's not clear to me how this would be managed using other proposals, as throughput in a logistics network is probabilistic, depending on request size, distance travelled, charge port location, and current network congestion. It seems to me you would have great difficulty in managing things in linked networks because the only thing you can set is request size. Splitting networks allows each to perform it's own logic locally and act accordingly. E.g. only share green circuits if there are more than 2k in the network, using a fast inserter to limit throughput.

Well, Factorio aims to be this:

A wired, confusing, experimental, mind-blowing, always-something-todo puzzle game.

Compare that with your screenshoot.

Again, seems a little unfair to compare a deliberate mess of spaghetti, with a deliberately simplified demonstration
I submit instead a picture of my factory where I have experimented with this idea:
Not quite the same level of tangled mess, but I like it. (Also far too many ports -- as I said there's a lot of experimenting in this map)

And I can assure you there is no fundamental problem with scaling this concept

I've actually expanded a decent amount further north than this, but the game didn't seem to like zooming out to 0.0125; I didn't want to push it further. Controlling bot count becomes more difficult the more ports you add, but adding a chest to each is really not a prohibitive cost by the time you have roboports in that kind of quantity.

To be clear, I don't claim my system is perfect either -- there are many things that are unnecessarily difficult or impossible currently. But I feel that most current suggestions add a large degree of complexity compared to the increase in functionality, and simultaneously create a new set of issues that will have to be dealt with.

[ssilk]

Hm. I try to bring this back to reality.

First what you propose with the connected networks is an interesting idea. No question. But it's nothing which must be suggested, it's already possible to play like so. Or I miss here something. I think what you really should do is to go into Show your Creations board and explain your factory in detail.

Second, the original suggestion was about construction chests. I don't see them anymore. Seems like the suggestion has changed to a new topic. But I have no idea what topic. So if you want to continue with discussing this idea as a suggestion, I want you to begin an new thread first.

Third what you propose with the logistic connections to different networks is either a completely different game than Factorio, or it's the top of an iceberg for a cool addition. And here I point again to the suggestion with the overlapping logistics II. I admit: Difficult to read. (You asked if the robots are shared: yes, they are.) He point of the overlapping logistics is nothing else than to able to have logistic chests in different networks. You can for example define a repair network, by having requester chests requesting repair kit and walls and then insert that into passive providers which are connected to the repair network. That forbids for the logistic bots to put repair kit from the passive providers back to the requester, cause the requester is in the default network and the provider in the repair network, but allows a construction bot searching for repair kits to use it. In short: You can create complex networks without the need to separate the roboports.

This is different to your idea: you want to create a different game than Factorio. You want that the players change their behavior, to play a different style. That won't happen.

[TI-89]

Oh, I don't expect anyone to build what I did - That's just me. Networks in general are a fascination of mine; plus I found it easier to start reasoning about how to distribute bots if I knew there were enough of them to go around. So I just massively overproduced.
But, now that I've figured out most (I think) of how this system works, I'm trying to figure out how to show this to other people. Not because I want them to change how they play, but because while searching forums trying to figure out how the hell roboports are supposed to work I saw a lot of people posting the same questions I had. I did get a bit away from the construction chests but what I was trying to explain with connected networks is that there's no need for such a mechanic because the functionality already exists.
And you don't have to take it to the extreme I did; that was the point of the 'toy problem' -- it works on all scales. And the same is true of overlapping logistics.

You can for example define a repair network, by having requester chests requesting repair kit and walls and then insert that into passive providers which are connected to the repair network. That forbids for the logistic bots to put repair kit from the passive providers back to the requester, cause the requester is in the default network and the provider in the repair network, but allows a construction bot searching for repair kits to use it.

But you can already do this, by separating the repair network. And besides getting rid of the feedback loop, this also ensures that the repair packs are nearby when a construction bot looks for them. Because with overlapping you could still have bots fly across your base as biters chew through your walls.

But yeah I'm working on the separate thread thing. Arguing with people like you (e.g. those who actually read through my long ass posts and criticize) helps me gather my thoughts and lay them out. Otherwise an in-depth explanation of my factory would be incomprehensible, and as long as War and Peace . So again, and sincerely, thank you.

Plus at this point it's mostly just a ton of roboports. I think I've almost got trains figured out, and I'm starting to scale up and decentralize production. Then I post

[Jürgen Erhard]

This actually needs a bit of revisting, since in 0.15, as we all know, those additional science packs (and the old ones too) do use even buildings.

And I find construction bots do not get equal priority with requester chests: if a provider chests had item N, and multiple requester chests want it, they get some served in some kind of round-robin scheme. Construction bots/jobs apparently do not factor into this. A construction chest would solve it (I usually "fix" this by a simple circuit network… circuit regulating inserters into science assemblers based on supply of N: only insert if there are more than n N ).

[JohnyDL]

but 0.16 will have buffer chests so if you request them to assemblers you can steal them back out for construction, I'm kinda hoping this is just a "I didn't manage my factory well enough to have items for construction and science" rather than an actual problem, make miners in 2 places, science and say a mall and the science miners are on belts to make science and the mall miners are in the mall in chests for use.