What Sucks Nov 2022

[garrynewman]

I'm gonna start asking this more and more as we go on. Please tell us the top 3 things that suck from a development point of view and from a player point of view.

Developer

• Asset publisher is difficult
• Hammer compile times
• Sometimes need to restart editor

Gamer

• Construct performance
• Crash when changing map
• Assets broke when switching between games

[Darkedit-dev]

Dev:

1. The sometimes outdated and not so detailed wiki/documentation. :)
2. Unsupported css types. Going back and forth spending time on css that is not valid and finding alternatives (Is there a list of what css is supported?)
3. When there are errors in script runtime, you don't get good error feedback like what line and what script is causing the error. You get this in Unity which is very helpful for faster debuggin.

[Remscar]

Some additional documentation/wiki requests:

• Proper way/best practice to do a server side tick for gamemode (Game) code
• Variables / parameters of entities exposed to hammer?
• Mention event system in entity documentation (https://wiki.facepunch.com/sbox/EventSystem)
• No documentation on IUse or usable entities
• No documentation on how to create ui panels actually show up. (Figure out that you just new PanelClass() on the client, but it would still be helpful to have this explicitly stated somewhere)
• <h1> tag not supported in html? Would be nice to get a clear list of what's supported
• How to make the cursor show up to interact with UI / how to make interactable UI

https://wiki.facepunch.com/sbox/Entities - Mention event system in entity documentation https://wiki.facepunch.com/sbox/Linking_Entities_to_Hammer - Variables / parameters of entities exposed to hammer? https://wiki.facepunch.com/sbox/GameLoop - Proper way/best practice to do a server side tick for gamemode (Game) code

The documentation about IUse is available in your IDE and on https://asset.party/api/Sandbox.IUse, but I will add a mention about it on the wiki for visibility.

No documentation on how to create ui panels actually show up. (Figure out that you just new PanelClass() on the client, but it would still be helpful to have this explicitly stated somewhere) - https://wiki.facepunch.com/sbox/UI

How to make the cursor show up to interact with UI / how to make interactable UI - https://wiki.facepunch.com/sbox/pointer-events and https://wiki.facepunch.com/sbox/ui-events-and-input

Awesome thank you for those links and making some changes to the Entities wiki page, I'm sure that will be very helpful to others.

I still think the documentation in (https://wiki.facepunch.com/sbox/UI) is a bit lacking in how to actually use the ui. It does a great job telling you how to make your own ui, but leaves out that after you've created your class, all you have to do is create an instance of it on the client for it to show up.

I'm realizing that i'm able to make edits to the wiki myself, so I may go ahead and add a section about this.

[Remscar]

Can you please elaborate on specifically what documentation feels outdated or not detailed enough?

API docs are here: https://asset.party/api Tutorials are on the wiki: https://wiki.facepunch.com/sbox/

Some tutorials/helpers on NavMesh usage would be helpful.

I've been trying to get NavMesh.PathBuilder to work, and it does in most cases but it's hard to debug/diagnose issues and it occasionally stops working.

[robotboy655]

NavMesh stuff is on hold until a painday as there's a lot that's going to change about it.

[Remscar]

NavMesh stuff is on hold until a painday as there's a lot that's going to change about it.

Sounds good, if there's ever a discussion soliciting feedback or feature requests can you please @ me on it if I don't see it? There's a few feature deficits i've found in the current structure which I think could make it a lot easier to use.

[robotboy655]

Sounds good, if there's ever a discussion soliciting feedback or feature requests can you please @ me on it if I don't see it? There's a few feature deficits i've found in the current structure which I think could make it a lot easier to use.

You can stay tuned to the official Discord server announcements & commit log channels, or this Discussions section (https://github.com/orgs/sboxgame/discussions) for when the nav mesh changes will take place.

You are free and welcome to provide any feedback on the changes/system itself then.

[yuberee]

Dev: Ignoring the things that I know will be added in the future: The crashes
I feel like I get a crash every other compile.

[garrynewman]

Can you elaborate?

[yuberee]

Of course.
I've had S&box crash frequently, mostly while hotloading code in game. Compiling my game while I'm on the main menu doesn't really cause a crash, but depending on what I'm working on (Editing animated entities for example is more likely than editing gameplay elements) S&box freezes as soon as I run my game.

[Fortune117]

Developer

• No ServerRPCs.
  This is very annoying to work around when designing things like an inventory system where the client needs to send changes to the server all the time.
• Can't see error trace in play mode. Makes bug reports harder to follow up for difficult to reproduce errors encountered during regular play.
• No RPCs on entity components
  I use entity components all the time, and having RPCs on the components themselves would save me a lot of boilerplate code.

Gameplay

• Performance.
  It's really bad right now. I don't want to sugar coat it either, I have a 3090 and a good cpu and I still get sub 30fps on maps like Office and Construct, especially when in dev mode.

[GRAnimated]

I'm on a 2060 and I get 100-120fps anywhere in Construct, are you sure your game isn't using your integrated GPU?

[Fortune117]

I'm sure. The game is heavily CPU bottlenecked for me.

[Thomas-Lazenby]

Dev:

1. Networking.

• When doing anything with generics its broken, and have to do {ClassName}_{YourGeneric}. (But that is very specific).
• You don't get warned if you are an idiot and attempt to change value on client side for just a [Net] (I might be skitso and thinking this was a feature). (But that is very minor).
• ServerRPC, having to use ConCmd.Server is ok just makes code a bit more complex.
• Not a lot of commenting on NetRead and NetWrite.

2. Access Compile authority?

• When you compile it wont tell you the issue and fails to compile.

Gameplay:

1. Performance for servers

• Last playtest I did with 20 ppl it just lagged to death but might be okay now and crashing half the time.

[trende2001]

You can use dedicated servers on your PC (they re-route through the steam traffic meaning that your IP isn't exposed) for large playtests of up to 20 players. Also yes, instability, s&box has that right now since it's in dev-preview mode, not trying to sugar coat it

[samiscool51]

player point of view:

• screen recording is annoying since i didn't know what activated it and didn't know where it stored the vids. found out it was some random button on my g502 after slamming every button on my system, i also found the folder that contained the vids, it took up 20gbs of my harddrive space once i found it. fun.
• the console log isn't detailed enough so when something goes wrong i usually have to take guesses without much to go on, very annoying when i'm trying to figure out why something isn't working/loading.
• the game needs to have an in-game option to clear downloaded assets, the stuff can take up quite a lot of space and it's unintuitive to the user to go to their sbox install directory then clear out the assets folder manually

and as much as i would like to talk about performance issues on older GPU's, it's been discussed to death so i'm not gonna go into it.
it's been frustrating dealing with sbox, it acts like a toddler, getting mad at the smallest thing and not telling WHY it got mad.

i do hope these points help you improve the game garry.

[tjones4701]

DEV

1. Addon code/asset dependencies - I have created code before that i want to reuse and in order to do this i have had to copy and paste the assets/code folders into new projects. Being able to just 'release' it and then use them as a dependency would make it 100x better.
2. HTTP Clients Missing
3. Hard to create a dynamic mesh/model using code. - I want to create a simple box.sphere etc.. for prototyping. It's harder than it should be.

Gamer

1. Havn't Played Enough Yet

[NotGaylien]

Dev (artist):

All my main dev issues have been around the c# asset browser, forcing myself to use it over the old one for while or so now to find out the main issues im having with it.

• the search bar feels very clunky and painful to use. 90% of the time i try to search for "table" and look up and either its done it in the middle of the file path, at the end of the file path messing it up and making me have to delete it if im lucky or find the whole file path all over again just to try find the table model in the addon i want.
  i feel like it would be a lot better if the file path wasn't there since it has only cause me constant issues and i have had 0 use of it being there, don't think a single time i have typed out or changed a file path other than when i have accidently deleted part of it while trying to search for an asset.
• Also general navigation of it just doesn't feel nice, i always feel like im clicking through different files back and fourth to access stuff, like if i have 1 addon with certain models and another addon with others, i have to go back and fourth between the both constantly rather than just being able to have them both access as one, it feels like a file browser with extra sbox stuff on top and not an asset browser its just really off to use when i want to manage, mess with and browser assets as a whole, and not individual files.
• asset type filtering becomes a mess in the search bar real quick, when im filtering for more than a few asset types the search bar gets instantly cluttered, for example i like having : model, material, image, particle, animation graph, fbx and some custom resources.
  you can see how suddenly my search bar looks like "t:vmat t:clothing t:vpcf t:fbx t:vanmgrph t:jpg t:vmdl t:decal e:/sbox-games/spookyjam22/ table"
  also i have accidently removed them filters while searching for an asset and i have to go back in and add them all back so many times its painful, they just clutter the search bar so much also i have had random issues where sometimes the filter pops up in the middle of a filepath for no apparent reason and i have to go in and delete it out again.

i would say they are the main 3 issues i have had
TLDR

• filepath in searchbar hasn't been useful at all for my case as an artist have had no use for it being there it just has caused me issues.
• feels too much like a file browser to browser around in individual files or directories, and not a nice way to manage and deal with assets in sbox as a whole all or some addons.
• filters clutter searchbar and causes issues being there.

[gmodn]

Before I begin, I'm so happy that you guys actually care about something like this. It's so easy for companies to just ignore important feedback like this.

Devs:

The wiki documentation is very outdated in some areas.

Console log sometimes isn't very detailed.

Those are the only two that comes to my mind as a dev. You guys have done a really good job with the editor and how development goes in s&box.

Players:

There needs to be a way to access downloads of games and manage where all of that goes. Having a delete content option would really help too.

Gamemodes show up in the menu even if they're not working.

Hope this helps! It has been really crazy seeing this game get developed and it's honestly incredible what you guys have done the past two years with source 2. Keep up the amazing work.

[sndrec]

Dev

No ServerRpc.
asset.party API documentation isn't searchable.
Support for client authoritative gameplay in general is a little weak.

Player

Do not like the sausage people all that much. I think it's the face. Would rather a more human adjacent base if not just the ability to pick and choose playermodels like Tower Unite.
There are long stutters and freezes when loading new assets into memory - it was especially bad for me and a friend some few weeks ago where we were freezing so long we would time out, after shooting the regular pistol. This remains an issue for my monkey ball gamemode where players will freeze as a new stage/background is loaded.
I think the default sort for gamemodes should be random, starting with gamemodes that have players, followed by gamemodes that do not, rather than most popular/most players shown first, in order to incentivize players to experiment and try new modes.

[Gmod4phun]

Developer:

• Missing documentation on some interesting things.
  For example RenderAttributes, I couldn't find anything about the default attributes, to see what is and what isnt even possible to do with those at the moment

• Lack of some more advanced control about entity collissions.
  The current Collission Matrix system is pretty handy, but I feel like there is no easy way to quickly disable collissions between two specific entities on demand without affecting other ones. I'm not sure what would be the best way to do this, probably having some lists for each entity, for example AllowedColliders and DeniedColliders which hold other ent references, and they would have priority over the Collission Matrix. Just food for thought.

• Lack of more advanced entity rendering.
  To be specific, a quick way of using Render Clip Planes to visually clip individual entities on demand

Player:

• Performance.
  Used to be alot worse, its better now but I believe it could be alot better
• Some form of cleanup of gamemodes/addons
  After playing for a while, the sbox folder gets bigger and bigger. It would be nice to have some sort of a cleanup button to get rid of unwanted downloaded content

[SupinePandora43]

Last time i used s&box there was "NoCollide" constraint

[P0stf1x]

Dev:

• As a starting game developer, I feel like networking is really unintuitive and there's almost no documentation/examples on how to do it right. I know it's still closed beta and things change constantly, but still, there are very few things said about it on the wiki
• (Almost) Every time there's an update in steam, the editor will crush when hotloading/compiling code

[GRAnimated]

Developer

Graphic glitches on integrated GPU
UI problems (things keep getting moved in the editor/hammer, not documented) (hard when following S2 tutorials)
Uploading to asset.party is confusing

Gamer

Not enough players
Games using private maps that are not uploaded are shown next to public games
File size bloat

[JohnyBro]

Developer (new to game dev)

1. Networking is hard to debug (for example prediction errors aren't detailed enough)
2. Some part of the wiki isn't up to date and need more examples

Gamer (didn't play that much)

1. Can't add games as favorite to find them faster
2. Not a big fan of the sausage man
3. Performances

[DoctorGurke]

you can actually add games as favorites to make them easier to find

[JohnyBro]

And after that where can I find the list of my favorite gamemodes ?

[DoctorGurke]

at the top of the main menu as smaller icons

[Jammie3d]

Dev:
Learning how to Create & edit shaders is very daunting to a non-coder/dumb artist like me.
As far as I know we dont have access to all of the uncompiled shader filles to look at and learn from.
We also seem to be missing things like Voronoi noise, which in other software I would be able to add in as a node and quickly utilise as a mask for a lerp or something.
In an ideal world sbox would have a node based shader graph but I understand that this is a big ask, especially before launch.

(I may be wrong in some of these cases because I have only tried to scratch the surface of shaders so far.)

[Redhacker1]

The thing about It is You're not dealing with a shadergraph. Most other engines give you a limited shadergraph, sandbox exposes a whole shading language which is significantly more powerful, but it also does mean you probably need to learn some coding and more responsibility is placed on you to implement the features you want. I think once things start moving along we might start seeing more shader graph type setups being provided by the community.

[gio3k]

Developer:

(Keep in mind that these issues come from me wanting to use s&box as a game engine. I also don't expect these to be fixed instantly, keep reading to see what I mean.)

• No (as of right now) standalone game support
  The biggest feeling I get right now from s&box dev. is... "ok, but is it a good idea to keep going?" which certainly isn't the nicest feeling in the world.
  I love the engine. I want to make a good game that could take actual YEARS of development...
  I don't want to have the problem that sometimes happens with Roblox etc. where people talk about a really good Roblox game like it's "impressive that they thought to do this in Roblox" or "a good Roblox game"
  I'm not saying the above is a bad thing (I feel like it's even a great thing for a lot of the current s&box games), but it's really not what I want if I'm making a long singleplayer game over multiple years.

• Not great / non-existent cross-platform support
  This is my biggest problem when combined with the previous issue.
  I don't even mean macOS or Linux, I mean devices like consoles, maybe even mobile devices.
  It's sad, but I don't have much hope for this to be addressed.

For both of these, the responsibility should of course be put on the game developer if they want to take either route. It shouldn't be up to Facepunch to keep a standalone game playable, bug-free and optimized (this is a big one for cross-platform)

Again, the feeling I have comes down to this: outside of the s&box "bubble", if I spend 2+ years on a Windows-only singleplayer RPG that's stuck in the s&box "game", then I'm just an idiot. No way around that.

I love the engine and all the work that's been put into it to make development fun and interesting... but I really need to have some insight into the future of s&box before I can put my all into this game.

[garrynewman]

I agree 100%

[DoctorGurke]

P.S. Game Templates pls and deal with base properly :( poor obsolete folder

[xezno]

Dev

• The workflow for uploading something is weird: I make my project in the editor, I upload it in the editor, I change some stuff - title, maps, tags, category - in the editor.. but I then have to switch to asset.party to upload thumbnails or screenshots? Why can't all of this be in one place?
• Editing a map and previewing those changes takes far too long. Even a fast compile can take a few minutes, baking lighting takes an actual eternity.
• I want more game templates.. a lot of base stuff got obsoleted, so I don't want to use that, but I also want to be able to start with something that isn't just a flying melon

Gamer

• The menu is really inconsistent in a lot of places (both UI and UX wise)
• Everyone's mentioned this 1000x but I'm echoing it because it's a big one: performance sucks.. I'm on a 3080ti and construct runs at like 50fps. This is worse than any other game.
• Server list has no filtering or search options, when the game launches in 1-2 months time or whatever there's going to be a billion servers and no way to filter through this, making it useless

[matekdev]

Developer

Networking needs a complete rework in my opinion. The base system is simple and easy to use, however, there are some missing features.

• "ServerRPC" or whatever you want to call them
• The ability to choose which data you want to network to X players, right now everything gets networked to everyone with no real choice (this should also include components).
• BaseNetworkables have a bunch of missing features (for instance network ident doesn't even work for it), most of the time people just say use an Entity instead of a BaseNetworkable.
• Networking during hotreloads almost always break
• Network types don't work properly, i.e networked dictionaries, networked list, etc. half of these are either broken or don't work at all once you start using multiple levels of them
• ClientRPCs also end up failing to send properly when players have higher pings

Player

• Performance is meh as other people have mentioned

[emesare]

Glad you mentioned the ability to choose what to network to x players. Here is the accompanying issue: #2108

[Baik90]

Dev:

• Some Missing Entity´s like point_template to respawn another entity
• In general, the Map Input output feels Outdated.
• Ether I am blind or you cant change the Material group ingame via the Inspector.

Player: Nothing

[DoctorGurke]

the inspector asset preview is extremely limited right now, it sucks. There's a big issue up for it

[Volantarius]

Developer

• Handling tons of assets sucks butt. Really want a batch script convert tons of textures to get them to stay consistent. I've made a bunch of wall textures and they need to have the same specularity, metalness, etc and its such a pain to do it by hand. Writing python scripts or anything outside of the engine is shit, and sbox would be where I would like this to happen.
• Quiet, really quiet. For developers this should be super simple to place tons of entites similar to ambient_generic's and have hammer create a soundscape file for the level. In hammer preview for how the sounds will work and to create the triggers to change DSP.
• Tips and help in dev mode. The wiki should hold all of the solutions and issues one would come by, so that users can add and edit the wiki to keep up to date with everything. C# errors could also be pointed to the wiki to have a better solution than reading documentation or searching forum posts again. And even better the wiki can have the links to forums posts or documentation which can all be annotated.

Gamer

• Gamemodes being deprecated or wrong version. Numerous gamemodes I've tried playing are old, or the back end is shut off, yet are still visable on the menu. Having a tip saying the gamemode is old, or a ping to the back end to check if its up can probably help.

[rileyzzz]

Developer

• Procedural stuff still kinda sucks. We can set up mesh data and bone transforms and collision, but after that, not much. We can't add attachments, we can't set up ragdolls, we can't do IK, etc. Not sure we can create surfaces at runtime to use on the models either. No matter what gets added to ModelBuilder, there's always going to be something you can do in ModelDoc that you can't do with a procedural model - I'm assuming they're handled completely differently by the engine.
  And in practice, it seems like procedural physics mesh/hull collisions and traces can be a lot slower performance-wise than baked map physics - even if they're static. This could either be completely placebo or a rubikon limitation, I haven't done any testing - but if it is there, what can be done to close that gap?

  • Ultimately the best way to fix these would be to keep pushing to move as much as possible out of C++/internal glue code and into C#.
    Does ModelEntity really need to be an engine thing? What's stopping it from extending Entity in C# and using SceneObjects - probably mostly just instancing, physics, and animation.
  • Instancing - devs might want to do this with sceneobjects anyway. right now it's entirely engine controlled (and usually turned on), which causes issues with render attributes.
  • Physics - Trace.Ignore takes an Entity parameter - why not a physics body/group? SetupPhysicsFrom* exists in ModelEntity - why not Entity? And why can't an entity's attached physics body be set/handled in C#, besides just accessing it?

• Despite how neat the SceneModel, Trace, and shader APIs are, Entity is still all over the place. It's chock full of garbage that doesn't need to be there, which inevitably causes friction and confusion when these things get removed piecemeal. IIRC, a while ago Garry had said that Entity should basically just be a Name, a local Transform, and maybe some parent data. If that's still the goal, this needs to be dealt with ASAP before release.

  • EyePosition? GroundEntity? Health? WaterLevel? If these are still important to the engine, they need to not be. But, and I can't stress this enough, don't just remove the functionality. If Entity.EnableShadowCasting gets removed, there needs to be an intuitive way to do the exact same thing in C#. Same goes for ModelEntity.
  • Lean on components a bit more, instead of just using it for glow and other minor crap. Evaluate every entity-centric system and see if they can be reimplemented as a component. Entity IO? Prediction? its physics bodies? following its physics bodies (i.e. PhysicsMotionType.Dynamic)? just spitballing here.
  • That could be the biggest pain day ever, but if it's going to happen, I think it's worth doing before release to save future headaches.

• The shader API is still somewhat limiting. Where are storage/structured buffers? I have to use RW textures to manage large data. Why can't we queue buffer/texture copies? I had to use a compute shader to do that. Why is it ass to upload a cbuffer that uses an array? Why can't we set combos directly on materials? The workaround to setting explicit material combos in C# is attributes, but they break as soon as the object gets instanced... unless you're using dynamic expressions.

  • It would be insanely useful to have an equivalent of the C# api page on asset.party for shaders. Especially since we don't have intellisense. Right now, we either need to go digging through the base/shaders/common folder, or know some magic function names from the unexposed .fxc files. Finding helper functions like world space conversions are a pain in the ass and I usually just end up doing them myself.

Gamer

• The discord server is way too unfriendly to new players. This isn't really facepunch's fault, and there probably isn't much anyone can do about it, since most discord servers these days are pretty toxic. But that can be a pretty major turn-off for anyone trying to get involved in the community, and especially for anyone who joins trying to start out in gamedev. Not a big issue right now as it's still technically a closed beta, but just throwing it out there - getting some more moderators around release time might be a good idea
• The ui design changes like every 3 months. I'm not a UI designer, and facepunch has made it clear that the UI is heavily subject-to-change, so take my words with a grain of salt. That being said, right now the main menu looks like a mobile phone. I'm starting to dislike how limiting google material icons are for s&box's use case - does it really have to be that minimal? I feel like we need some extra, more s&box specific facepunch-made icons to work on top of material icons.
  • Take, for instance, the current icon for the avatar editor: emoji_people. It's a guy waving - what does that mean at first glance? Friends list? Social/facebook/twitter/discord links? 'Avatar Editor' is probably one of the last things that comes to mind
    Same for 'Streaming Setup' - what is that? Icon makes it look like a videos feed (i.e. steam community), or maybe a link to facepunch's youtube or something - nobody's ever going to click that.
  • Then we have three games lists. 'Home' - shows a really big list of popular games, then there's 'Game List', which shows... more games, but it lets you sort and search them, then there's 'Local Games', which shows the games you have locally, but to add stuff here, you have to exit the application, open sbox-dev, and manually download and 'enable' them. I feel like this could all be one thing. If you really want game previews in 'Home', only show like the 3 most popular games, and make it so that clicking them instantly puts you in - kind of like a quick-launch. I usually hate quick launch buttons, but I could see people using it if they only ever open s&box to play one game, like TF2:S2 or something.
• Artstyle is all over the place. That's kind of a good thing, gives people a diverse set of themes they can shoot for with the games they make. But if that's the case, why don't we have differently styled playermodels to match? I'm not making the 'sausage man looks bad' argument here, definitely keep Terry. But if he's supposed to be a 'base' character, that all games can use, why not have different styles? Can't we have a cel-shaded, comic book terry? a realistic, grungy Terry? How about a 2d terry that we can pixelate in post-processing, for arcade-like games - could that be made to work with animgraph? The Terry we have right now is still very first person shooter-centric.

[trundlr]

I definitely disagree about combining the games lists into one. The home page is obviously going to have work in the future, and so will the 'Game List', in order to promote discoverabilty. But the local addons tab is an easy to get to tab that a lot of s&box players will never need to use - having it on the home page would take up extra space that could be dedicated for other things. Personally, I trust Facepunch in the direction of this, since they have may hold a grander vision of what they want s&box to look like in the future anyway. Just my two cents 😸

[Geeknerd1337]

Dev:

• Despite the many improvements made to audio, there's still a good level attention that I feel should be applied to it. Right now it's difficult to make a game revolving around sound and the timing of sound. You have to do weird hacks with things like music because if you play them when the game loads they mute themselves after about two seconds. You have to turn prediction off every time you use a sound from the server side and I feel like this should be automatic then as I haven't found a use case for playing predicted sounds.
• Shaders can be frustrating to work in sometimes. While the underlying HLSL isn't difficult to work with, having some kind of more verbose shader reference in the documentation would be nice. There are times where I feel like I'm just trying random settings to see if I can get a cube map texture to show up or whatever and even with the improvements to shader compilation, it can take 30 seconds to 1 minute to see those results in game.
• @lotuspar put it best, but standalone game support would be really nice. I feel like I've put a lot of time into s&box, created things that I feel could be legitimately decent IP outside of the ecosystem. I consider s&box to be one of the best game engines I've ever worked in and at the moment I feel somewhat limited in a way where the only way I can hope to make money off of it is to either work for a more wealthy team or hope that my game is engaging enough and ticks the right boxes to be popular within the s&box menu.

Player:

• Performance, it feels like the games performance has steadily declined for me over a year. Not just graphics, but networking as well. If you play any somewhat advanced game with more than 8 players, things begin to break down. I know there are ways to mitigate this with custom networking and hosting dedicated servers, but if the vanilla experience of s&box is intended to primarily be P2P "I host a game and my friends join" then perhaps there's room for improvement.
• Games menu icons can be a little confusing. It's hard to tell what button does what and even as a dev I find myself forgetting what and how to do things. Additionally, sometimes the menu just runs into issues? Server lists don't seem to update properly, games appear mostly empty then say they're full when I try to join. It's mostly just bugs to be ironed out but that's my experience at this point.
• The UI for terry's clothing is a little confusing at the moment. I know it's in progress, but there needs to be a way to tell when selecting and item is going to remove other articles of clothing. Some stuff can be worn with other stuff and others can't and there's no real indication of what can be added on to vs what can't. I like fiddling around with my Terry and stylizing him, but that small hiccup makes it just a little less fun.

[Mike466-0]

Dev:

• Importing Assets Workflow. Putting all the assets together to create a model/material that is useable takes up alot of time when you import large amounts of assets.
• New Asset Browser is slow to load and the search function is confusing. Example, "@recent" or "@cloud" would rather have a drop down instead of having this in the search bar

[peter-r-g]

Dev:
I'm a big fan of networking so my complaints will largely revolve around it.

• Extensibility of networking types. Right now if you want to create something that is networkable and not supported by S&box out of the box. Your only options are a BaseNetworkable using INetworkSerializer or using a ClientRpc/ServerCmd to transmit. The former is limiting users exclusively to the BaseNetworkable class and does not support any form of delta changes. The latter can work Server > Client with the limitation of being forced to use an Entity. In Client > Server, it is very difficult or potentially impossible as no binary formats would be supported by ServerCmd. My potential solution is one that I have implemented myself in my own custom networking solution that works with S&box through web sockets. The idea is to use an interface that defines something that can be serialized/deserialized both whole and partial. You can view my implementation with tons of documentation here if needed. Using this approach allows me to make any simple or complex type that I may ever need. Items include Vector3s, generic dictionaries, or even implementations of base networkables/entities.
• ServerRpcs, I feel at this point mentioning this is like beating a very dead horse but it is still a very useful tool to developers. Being able to fire (and maybe forget) a bit of data to the server is extremely useful. While it is possible to cover most use cases through a ServerCmd there are of course issues with it. Firstly, it is a command. A command that is open to users to mess with any time they please. Whether intentional or accidental on their side it poses an unnecessary problem and headache to developers that they now need to make sure players aren't stupid enough to mess with it. Additionally, with it being a command, it is impossible to send any binary data to the server. Instead, you'll have to come up with your own representation in a string that also does not break the size limit on commands.
• CSharp projects, being unable to make any of your own changes to them can be very frustrating. Especially if your project may venture outside of the S&box game itself. Examples include but are not limited to: Being unable to reference other projects, use analyzers, remove the global use of Sandbox.Internal.GlobalGameNamespace, etc are all issues I have to consider when working on my custom networking solution NetBolt. To tackle the first issue, right now I have a .bat file to copy all shared code then gitignore that client version of the code. But with that solution, I now have the problem of exposing all internals to end developers. While I can put a notice to not use internal functionality it's still visible and a developer may not realize it is internal. The second point I have not had to directly tackle yet as the project hasn't gotten that far but I plan on using analyzers to make sure developers aren't doing something stupid like using a ServerOnly piece of code in a client-side context. With S&box, it is currently impossible to do that due to no support for inputting analyzers or editing the csproj. With the global using entry, I am a person that very much prefers using exactly what I need and not bringing in any functionality that I do not need. Having the global game namespace visible everywhere is just an annoyance and would be nice if I could just turn that off and bring it in when I want to.

Player:

• Here I come beating another very dead horse... Performance and general stability. With an i7-8700k @ 3.70GHz and an RTX 2070 super, I don't expect my PC to become a jet engine when I load into a map like Construct. The problem becomes even more apparent when you jump into VR with Meta Quest 2. With my setup, the game is completely unplayable if I have tiled rendering on. If I leave it on, all 8GBs of my VRAM get consumed and S&box shits the bed not knowing what to do. Disabling tiled rendering puts me just under my 8GB budget but then there'll be many visual bugs like entities randomly disappearing at a distance.
• A way to clean up downloaded files. Whether it be automatic or manual doesn't matter to me. But at least have some way in-game to see what has been downloaded and from where with the option to clean some or all of it up.
• A way to ignore games in the browser. I don't care to see "Sandbox with X" five times.

[PixelQubed]

prop_physics needs animation support or prop_dynamic needs to come back!

prop_dynamic could do this. Now, prop_dynamic is gone, and there is nothing we can do about it.
Yeah, I could make my own entity, but that's beside the point because I want to upload this as an asset to asset.party and not have to upload an entire gamemode just to accommodate this one prop. It's anger-inducing and nonsensical.

But yeah, if you're all about consolidation Facepunch, shouldn't prop_physics and prop_animated be combined to make prop_dynamic so that we can get the best of both worlds at the same time?
If that's not possible, then maybe making an in-between entity or something would be beneficial, or adding the functionality to one and not the other to make either one of the entities redundant or make both of them pretty much the same?

I don't know, it's up to you, but from where I am sitting, all I'm seeing is; prop_physics + prop_animated = prop_dynamic

[robotboy655]

prop_animated replaces prop_dynamic.

prop_dynamic never did have physics simulation (gravity), so I don't know where prop_physics comes in in your examples.

[PixelQubed]

Well, we need physics simulation on some sort of prop entity with animation anyway.

[DoctorGurke]

Have you considered making an issue

[PixelQubed]

Have you considered making an issue

I guess I should make an issue and clarify what I'm trying to get across. I just don't like making issues for what others may consider "useless" or "stupid"

[SryEyes]

Developer

• Needing to reinvent the wheel. Especially when coding you need to write the same code for basic sytems again. This is time consuming, the foundation for lots of errors, bugs and performance issues. I know that the base addon code is desinged to be minimal and is good for people creating new and unique stuff but maybe different branches of example code would be cool. When creating a new project one could choose different stuff like a 2d or 3d base. With add in options like a default spectator system, inventory system, extensive weapon system and so on.
• Multirun / Server testing. The Launch configs are nice but testing server stuff or with multiple clients is way to laborious and should be integrated more into the actual envoirement.
• Performance Profiling. The Performance Window in the editor needs way more information about why something is slow.

Player

• Netcode / Prediction / Hitreg sucks ass. I don't know if nearly every dev and myself are to dumb to use prediction correctly but hitting someone is a coinflip when you arent the host. In most stuff like boomer i hit more than double the shots when im hosting the game.
• Joining and finding Lobbys / Server also sucks ass. Not being able to connect beacuse of a missing map, version difference, some gamemode code error, because its a private lobby is annoying and seeing all the unjoinable lobbys is also messed up. Also most data is way too old and there is no way to manually refresh. Server browser needs a bunch of filters and the home screen should show something different since right now its just the games tab with less options.
• Some way to customize your addon page. Most people right now use their about entry to list developers / give credits, list commands, announce changes, post their issue tracker or personal information and so on. Would be cool to seperate these points into custom entrys since when you want to get a specific inforamtion you ll need to read the whole text. I would like to just look at the page and receive the info i searched for in a second. Being able to shelude playtests or other events this way would also be much better than through a discord channel.

[TehFlaminTaco]

Developer

• Adding Models and such is a dreadful experience. Especially animated ones. Having to rebuild it in the Model Viewer really kills the workflow.

[gvarados1]

models with tons of animations take forever to compile

[gvarados1]

Developer

• Hammer issues
  • baked volumetric fog doesn't work
  • issues with blend & foliage shaders
    • no animated foliage and I think it still lowers fps
    • blend shader is the wrong color in-game
    • issues related to vertex color blending
  • default map template is awful
  • no terrain tools
  • no systems for 3d grass and foliage
  • baked rt lighting preview is inaccurate, doesn't account for postprocessing and some shaders are completely white
• New asset browser sucks (search bar doesn't work properly, ect.)
• Having to juggle between s&box editor and asset.party to change addon settings

Gamer

• Awful fps
• No per-game bindings. Sometimes I want to swap bindings for only one game
  • ex. In Corewars Duck & Shift are flipped to be more like Minecraft. I don't use the flipped keys in Minecraft so it throws me off.
  • ex. Some games have Pistol on slot1 and Primary on slot2. Sometimes I want to switch those
• File bloat

[damienfamed75]

Developer

• When you're in the main menu there's no way click to compile your code from the editor without launching the game
  • A few updates ago it used to compile even on the main menu, but now in order to update entities in Hammer currently you have to either keep restarting the game or be in the game and hot-load compile
• This was already mentioned but having the ability to use HTTP endpoints would open the doors so much for creating an organized backend for certain games that don't require a web socket connection
• Particle orientation/rotation is all kinds of fucked up and confusing. There should be a more straightforward way to control the rotation of a particle (sprite/model) than what is currently there
• The Particle Editor also crashes s&box very frequently whenever in-game and sometimes out of game

[ShadowBrian]

Developer

• I've noticed my game slows down from multiple subsequent hotloads, I'm unsure why but it seems consistent between multiple of my games. Only way to fix it is to restart s&box entirely... This is especially noticeable in VR where you basically live or die based on your fps.
• Speaking of VR I could use a "Start in Editor Mode in VR" shortcut in the startup shortcut list- I currently have my own sbox-dev non-steam shortcut for that...