(0.15.19) Blueprinted Big Electric Pole misses auto-connect

[yblondinca]

Good morning,

I am running a 2 player game with my friend and experienced the following suspected bug last night. I already checked to see if this was related to an existing bug report for blueprints not saving manual power pole connections in the "Wont Fix" folder. This bug is related to the automatic copper cable connections that should result after placing a blueprint.

When placing blueprint setup over an area that has existing Big Electric Pole connections, some of the blueprint Big Electric Pole automatic copper cable connections do not work resulting in some of my Small Electric Poles not being connected to my main power grid. Even trying to manually place Small Electric Poles near the blueprinted Big Electric Pole afterwards does not automatically perform the copper cable connections as they should. I was able to make the Small Electric Poles connect to the other blueprinted Small Electric Poles but not the Big Electric Pole for some reason.

I am at work right now but I will try to replicate the bug this evening and attach some screenshots to better explain the situation.

Thanks!

[Loewchen]

Sounds like one of the poles has reached its connection limit. Post the save please.

[mrvn]

It would be nice if electric poles would auto connect to both poles A and B only when A and B are not connected directly already. Maybe not even when they are connected over a third pole C. Although then removing C should reconnect A and B over a different third pole again, which might be too complex.

[Aeternus]

Poles will already not automatically connect to 2 poles that are connected to eachother. Poles have a maximum of 5 connections with copper wires (but red and green circuit allows for more then that). I've seen some wonky behaviour with power poles myself when using them at the edges of a grid of solar panels or accumulators - they don't always connect properly if they are in range of more then 5 other poles. Can't be helped though, easiest fix is to adjust your repeatable blueprints with this in mind.

[mrvn]

I've seem worse when doing laser turret creeping. And while I can't make a test setup right now I'm pretty sure I've seen plenty of poles connect in a triangle formation. A to B, B to C, C to A. According to you that should not happen and one of those should not auto connect.

[Rseding91]

Yes please post the save file and steps to reproduce. Although as others have said it's most likely working as intended.

[mrvn]

The examples I have when the wiring breaks down are from turret creeps. So tons of turrets in a cluster and it's real hard to see all the wires. So maybe it really never connects a triangle and it just looks that way because one wire goes through another pole without connecting.

But I've been thinking. If a pole is added and it doesn't connect to two poles that are already connected then that is not enough. It could also not connect to two poles that could be connected. For example you have poles A-B-C in a cluster. A and C are not connected because A and B already are. But when you add pole D it can connect to A and C. It should only connect to one since A and C could have a wire. I think this would reduce the number of wires used in tight clusters a lot and prevent exhausting the number of connections.

[yblondinca]

Here is a copy of my savefile right after the problem occurs:

https://drive.google.com/open?id=0B9mH3 ... HRnLVhnd0k

I have also attached my logfile

factorio-current.log

Logfile

(7.96 KiB) Downloaded 130 times

Here is the blueprint string for the radar outpost that I was building:

Blueprint String.txt

Radar outpost blueprint string

(893 Bytes) Downloaded 122 times

And now the screenshots

Here is the setup when it is working as a radar outpost:

working

Working.jpg (834.95 KiB) Viewed 5467 times

Here is the mining base where I intended to replicate the radar outpost:

mining base

Mining Base.jpg (866.27 KiB) Viewed 5467 times

Here is what happens when I place the blueprint down:

not working

NotWorking.jpg (979.29 KiB) Viewed 5467 times

Note the missing connections (Plus there is technically one more connection missing from a Big Electric Pole to the SouthEast):

missing connections

MissingConnections.jpg (172.99 KiB) Viewed 5467 times

Now I realize from the above posts that this is definitely related to the fact that there seems to be a connection limit for Electric Poles. My question is why is that limit set as low as 5 and not bumped up to 8 or something? There is probably a good reason for the 5 connection limit but I can see many instances where, 6, 7, or even 8 connections to a pole should not be out of the question. Otherwise you end up in a situation like the one I am in now where I need to rearrange things to make it work. For your average user, this does not seem very intuitive or documented.

Either way, lemme know!

Thanks.

[Rseding91]

Now I realize from the above posts that this is definitely related to the fact that there seems to be a connection limit for Electric Poles. My question is why is that limit set as low as 5 and not bumped up to 8 or something? There is probably a good reason for the 5 connection limit but I can see many instances where, 6, 7, or even 8 connections to a pole should not be out of the question. Otherwise you end up in a situation like the one I am in now where I need to rearrange things to make it work. For your average user, this does not seem very intuitive or documented.

Because a number had to be picked and it was 5. Almost nobody notices so it seems like it's just fine as it is.

The technical reason why it's not unlimited:

The connections are stored in a std::array<> of size 5 for improved memory locality allowing them to be accessed without having to do additional memory lookups. The same system inserter filters use and why those are limited to a maximum of 5 as well.

[golfmiketango]

Now I realize from the above posts that this is definitely related to the fact that there seems to be a connection limit for Electric Poles. My question is why is that limit set as low as 5 and not bumped up to 8 or something? There is probably a good reason for the 5 connection limit but I can see many instances where, 6, 7, or even 8 connections to a pole should not be out of the question. Otherwise you end up in a situation like the one I am in now where I need to rearrange things to make it work. For your average user, this does not seem very intuitive or documented.

Because a number had to be picked and it was 5. Almost nobody notices so it seems like it's just fine as it is.

The technical reason why it's not unlimited:

The connections are stored in a std::array<> of size 5 for improved memory locality allowing them to be accessed without having to do additional memory lookups. The same system inserter filters use and why those are limited to a maximum of 5 as well.

In my early days of factorio I used to build military structure blueprints with lots of power poles thinking this would help ensure connectivity if they were slapped down haphazardly in the heat of battle. Ironically the opposite proved to be the case as the internal connections created in such blueprints would cause the structures to randomly connect or fail to connect based on robot build-order.

The inability to specify that a particular power pole should not be automatically connected obviously does start to become a problem for anyone wanting to use switches in blueprints. It means the switches must be placed at awkward distances if the build is going to be repeatable... actually, come to think of it, how does blueprinting switches even work at all, is there some special wiring exception made for them?

A power pole with a teeny-tiny connection area (or which would only be connected if explicitly wired up, and could preserve those connections in blueprints) would be helpful for that.

[yblondinca]

I completely agree that the number of connections should not be unlimited.

However, I can see many instances where having an automatic wire connection limit of only 5 would cause reasonable blueprint setups to fail. I would guess that for every player that actually reports this issue, there are another hundred or so that have come across it and tried to find an in-game work around instead of documenting their problem here .

As your amazing team is approaching the home stretch with regards to finishing this game for a 1.0 release, it seems like this is something you may want to consider addressing beforehand if it would not cause the game to take a significant performance hit from making the limit 8 for example. You could lower the instances of a random user running into this problem from maybe 1% to .001%. Based on many of the FFF blog entries I have seen over the past few years, your team tries to optimize and make the game as perfect as possible. This would be another opportunity to do so if you have the time for it.

Whatever you end up deciding, I want you guys to know how much I appreciate the amazing game here you have created (I have purchased about 10 copies for family and friends to get them involved) and I will still play and recommend it to all my friends in the future .

Thanks!

[mrvn]

The real problem is not the connection limit but that the connections differ when you put down a blueprint. After all it all works when you made the blueprint.

In the example above the radar outpost needs 4 connections internally. So it can have at most one connection to the outside. Imho a poor design because you usually would want to have 2 connections to the outside to extend the network.

But the game could try to make the new long distance pole have only one connection to the outside.

My suggestion to skip wires when a triangle could be formed would avoid the connection to the south I think. But not the connection to the north in this setup. Lets get more radical in cutting wires. Why should a pole connect automatically to 2 poles that are already in the same network? The network does not need any loops to function. Could make removing a pole expensive and maybe surprising. Keeping a network connected when a pole is removed would have to check across all poles in the network and maybe create a new connection far far away from where the pole was removed.

[yblondinca]

You can close this bug report now thanks to: https://factorio.com/blog/post/fff-379