itch.io

even after 3 years the seeker is still seeking the flame 

V 1.1 072022 

- added playlist items browsing support

- fixed local mp3 files tags

- fixed iOS platform Control Center + metadata

- also fixed too tight init/download restriction

V 1.3.1 122023

- improved handling of decoder thread when stopping/destroying

- MPEG decoder: fixed incorrect state for certain MPEG files and updated used NLayer library build

it makes sense e.g. for grid based, non physics collision driven levels/movement

currently for testing I generate level and construct crude navigation map based on rooms coordinates (adding some offset to get the middle of each room)), the player/agent can move backward/forward in navmap - i.e advance from one room to the next/previous one:

( - not to confuse with unity's navmesh )

without any structural level information the data is only in tilemap now  - if you don't generate rooms/elements with physics colliders

(tilemap itself _might_ be enough in my case though - if the rooms generated had always exits only to connected rooms)

maybe things like A* would better operate on some preprocessed data rather than on tilemap tiles too, not sure

hi, please feel free to use whatever suits you and the overall style the best :)

I am fine with e.g. simple List of coordinates - I am adding to the list as generator progresses with each room right now directly in sequence generator, which works for it - but now it occurred to me that for branching/more complex paths a more advanced structure would be needed (which debug transforms w RoomChain component don't capture either I believe, if I'm not overlooking something  / it also looks like the debug transforms in the editor are not instantiated in correct order right now - they're ordered by coordinates, not path progression - don't know if that was intentional or not)

Anyway, although the overall structural information of the level might be useful in the future, I don't need a generic solution right now -)

- cheers!

hi matt, great update!

room sequencer seems to be producing very consistent results

[ I've only added null check for currentRoomTileSpace in GeneratorRoomSequence.cs:Ln 188, and also added missing board generator profile in MiniRogue BranchingTunnel Only Example scene ]

One small feature request - for adding a way to retrieve the actual final generated chain (previously possible via for example BoardGenerator's roomChainRoomLocationsFilled) for navigation purposes (at least I didn't notice anything usable). It's not a problem to add it manually to e.g. BuildRoomSequence, but maybe 'official' interface for all generators would be more user friendly :)

- Hope this helps, cheers 

i was just restarting play mode when it failed to produce chain on the first try, this was the same when generating in edit mode [clear&generate button]

when it failed the displayed chain was very short: 1- 3 rooms max - this seems to be correct though since there _are_ rooms w one entrance/exit only and requested minimum was 5

(as a sidenote: the condition for picking the next room should be probably more advanced and not entirely random if room template with more exits exists and end (edit: minimum rooms) was not reached - to not run in these premature dead ends; - but moreover there's no guarantee that the generation has successfully finished currently - I think you're safe to loop/yield until it is instead of hard limit though :-), but note that I'm totally fine with current state and can check result in user code if necessary)

'400' is enough for everybody, until a kitten is burnt ^_^

Cheers!

Hi, don't worry, I'm prototyping other stuff, not sure what (and if) generator should be the best yet

Yes, I've tried to move the whole setup before the loop in BoardGenerator into separate function and called that inside the loop before RunGenerators() - this results in non consistent state when generation fails ( generationSucceeded == false ) such that the minimal number of rooms is not reached and generated tilemap layout does not correspond to generated board data ( I have player placed inside first room after generation which is off the rooms chain in that case ) - probably something's not cleared enough in next iteration, but I did not investigate further yet

(there's plenty of space to run w 6x6 tiles and 102x102 tilemap, I've since deleted test package from OP, but I'll leave it here now: https://www.dropbox.com/s/tlp7sklzcu6i6fe/STroomchain.unitypackage?dl=0  (few SOs + scene))

I haven't tried other generators such as PlatformerRoomChain - I wanted very simple setup first with short (5-20), one-way non branching paths

it works neatly otherwise though :-)

Cheers!

So the problem is that if room chain generator is started more than once in single generation request - which it occasionally is since there's (hard) for loop in BoardGenerator.cs 's BuildLevel coroutine calling RunGenerators() checking its return value - you initialize starting conditions (location and start room) again - which is not in sync with whatever current data was being generated so far (including them in the 2nd run  would be probably incorrect anyway), so the generated layout breaks (despite chain being found correctly internally)

(I at first thought that the coroutine starting twice with combination with editor button click would be the problem, but 'decoroutining' it didn't help)

You might want to fix this properly - ( a quick fix directly in roomchain generator assigning initial location/room only on '1st' run is not stable, running out of iterations often. - The rooms seems to be generated OK if it passes, though, with Exists assignments being left as originally were)

Hope this helps, cheers!

hi,

i've made sure to be as careful as possible adding 15 (all combinations of four direction minus 1) room templates w simple tilemap (empty and non emtpty tile), but roomchain generator seems to be producing non connected paths often, 

- the Exits room lists for N and S seems to be swapped  (the custom inspector seems to be adding N to S, S to N, but E to east list and W to west list though) - note I've tried to manually swap room templates in N and S exit lists, but it looks like there's bug somewhere else (still getting non connected paths) 

and two things in the manual:

- there should be step 5.3.1 for adding the empty/black tile as default to board library

- picture in 5.1 is wrong :) - should highlight create new roomtemplate

do you want me to send a package w the project ? it's minimal reproducible after 1 - few random runs