Game Development - 2020-01-14

« first day (3449 days earlier) ← previous day next day → last day (1561 days later) »

5:31 AM

@trollingchar thats what im doing :D .

@TomTsagk yeah basics will do i think but i think its more of a other peoples style of coding that bothers or just my lack of skill.

@Vaillancourt, what do you think about c++20?

@Almo whats the intention of developers? Do they want more people to play their game? If about 100% of community dont like the changes why wouldnt you just give them what they want?

1 hour later…

7:08 AM

@Almo Being balanced for top players is not enough for a good game. It also needs to be entertaining for majority of players. Not all top players understand this.

I remember one RTS

main building was increasing limit by 3, so a normal strategy was to build 3 workers then a supply building

but there was a meme build

a player could build only two workers, rushing with one of them to the enemy base and building barracks there

not building supply allows him to do so because he is not spending resources on a supply building

he builds a single trooper and wins

unfortunately they abandoned the idea of allowing that tactic despite couter-tactics existed

(building a factory and when enemy trooper destroys your workers you can build a tank and strike back)

counter-counter tactic was to kill only one worker, forcing the opponent to build supply, slowing it down by attacking building under construction, while trying to build supply yourself while low on resources

it was so exciting

there were many different rush builds and counters to them

fun for mid-level players

I would like to develop a RTS some day. There are so many interesting things when it comes to balancing

3 hours later…

10:18 AM

RTS games are quite difficult to make, especially if you don't have any other code to re-use

they are very difficult to make

I always wanted to make some games similar to some warcraft 3 maps, but just getting the basics done is hard!

and real-time multiplayer is harder than turn based

@trollingchar i knew pvp game with evading mechanic. People could really abuse it to prevent heroes from jumping on them. But then i found out that heroes can still jump on them if they fall right onto them. I thought it was bug before but now i guess it was just hidden feature ? :D It was coded like that in UE physics i guess

@0x00004 reminds me of a game I used to play in the past, an MMO called Gunz (if I remember correctly). There was a mechanism where you can make a swift move forwards by double tapping the forward button, but it turns out if you do it on a wall while jumping, it refreshes your jump

People used it to climb anywhere, and everyone used it

10:30 AM

lol

@TomTsagk doesnt it sound like a bug?

It is a bug, but the devs wouldn't fix it, and literally everyone used it to attack each other from above

:D

So its more like if you didn't use it, you'd have a disadvantage

11:31 AM

If you had all the resources in the world, what kind of game would you make ?

there are many games I would make if I had all the resources in the world

and not having them will not stop me

it will just take a lot of time

True, but with enough programmers/artists, it would take less time

when I'll finish my current project, the next one will be a RTS

Oooh so you like big-scale projects?

@TomTsagk 100 programmers won't make a game 100 times faster

@TomTsagk yeah

11:43 AM

True, but 2 programmers can definitely do more than 1, if organised properly

I think you're right

I myself get easily bored if I work on a (personal) project for more than a few months

also if the project is modular, you can assign maintainers

There's plenty of ways to do it, all with cons and pros

It's difficult enough to make a maintainable project by yourself, imagine one with several people

it's difficult too

1 hour later…

1:13 PM

Something's strange on our devices. The query "update settings set state = 0;" executes successfully, but DB still stores previous value.

there are other issues with one particular device

like in this picture

user image

1 hour later…

2:35 PM

@0x00004 Haven't had time to check that much yet. I'll be limited by what MS has time to put in VS19 (eventually).

And now the query executed unsuccessfully, but worked

@trollingchar weeee!

When I try to execute the same query in pqAdmin, it freezes

I have figured out that's because the application holds the lock for settings table, but why other states don't cause it then?

3:11 PM

If I wanted to make a robot out of an hydraulic hexcavator, i.e. programmatically animating all the parts, is there a standard goto architecture for setting this up?

@0x00004 First, the community on the Forum is very small. We're talking on the order of a few thousand people, but the game has 2.5 million active users per month.

So there's no way "100% of the community" wants the old one.

And as trollingchar says, we can't just balance for the top few percent of the playerbase.

Old Ruin was absolutely horrible to play against until you can regularly hit great skill checks.

I can hit them about 50% of the time... and I have 950 hours in the game.

user92578

@Vaillancourt Yeah, at least currently support seems to be pretty lacking

And as I said in the thread, when playing the new one internally with a mix of new players (some of the artists and programmers aren't even gamers, so they're "new" players) and very high-skill players (5k hours or more), the new version is much more fun.

We've had this before. We change something, everyone hates it, a month later, they stop complaining and even start to like it.

With Freddy from Nightmare on Elm Street, he shipped and everyone went WTF IS THIS SHIT. and clamored for a rework

we started working on it... took like 1.5 months to do it. Forum posts started showing up "no don't change him! we get it now, and it's great!"

So these sorts of things are just a natural hazard of live game design.

3:29 PM

It's super difficult to tell in these situations. "There are 1000 people complaining in the forum but we have 2.5 million players so the complaints don't mean anything" is making it a bit too easy for yourself. From what I learned communities are great at identifying problems. If 1000 people complain then there is a legitimate reason why they are complaining. Also communities are terrible at suggesting fixes and usually misidentify the problems, so you have to investigate after them ...

... what they are talking about.

not that "they don't mean anything"

just that you look through the responses and when 10% totally agree with you and repeat your own design intentions without necessarily having read them, you probably hit it

if something releases and the complaints continue for months, then you have a problem

but in practice this doesn't happen that often to us

@Almo I think what @nwp is trying to say, is that not everyone is using the forums, and not everyone would complain if they didn't like something.

Having someone complain about something they legitimately didn't like is far better than users who just stop playing the game because of it

You already have a problem even it the change was perfect and people love it in a month. Maybe it's a trust issue because of past changes, maybe it's frustrations with other parts of the game that annoyed people to the point of just hating any change you put in front of them. Possibly it's a communication issue.

I don't think you understand the problem very well

have you been on the developer side of a live changing game with 2 million users?

it's very different on the dev side of that equation

you can't change anything without people being mad

we straight buffed a Survivor perk and Killers were annoyed because we "favor survivors"

but if we buff a killer it's all "god what don't you care about survivors"

That ... is just giving up. May as well do random changes, people will be mad either way, who cares.

3:34 PM

here's our communication 1 day before the changes shipped to the test realm

forum.deadbydaylight.com/en/discussion/117488/…

We don't do "Random changes"

Maybe you can instead collect survivor and killer nerfs/buffs in a package so it's not so one-sided. Small things like that matter.

we do deliberate changes, carefully tested with external players

we do surveys and data analysis on the responses

New Ruin showed "less frustration" for Survivors and "more fun" for Killers among the expert players we tested with

everyone on the design team agrees it's an improvement

this thing took literally months from design to release so we'd be sure to get it right

But still people are angry. But by analyzing their responses, we can see they have missed the point of the change, don't understand other systems in the game and how it relates, etc.

Which is great. And still you have a problem. The problem may be in the phrasing of the patch notes and has nothing to do with the changes. My point is you still have a problem that you should aim to fix.

this is a very complex, deliberate process and most players don't understand

we're working to improve with every release

there will always be people mad

you can't say "our goal is zero people mad!" because it's unattainable

our goal is "less people mad"

but when you change a perk used by 80% of the high rank players, tehre's gonna be blowback

LoL has been seeing this literally for a decade

Well, I can agree with "our goal is less mad people". "There will always be mad people" is just defeatist.

3:38 PM

but it's true

it's only defeatist if you let it stop your attempts to improve

which we do not

Sure it is. It's like saying software will never be bug-free. May as well write spaghetti code.

that a strawman arguemnt

It's not. It's the same argument as "players will always be mad".

I am not saying we should do nothing to improve. I am saying that we are doing things to improve while knowing we will never eliminate people being mad

yes it is, and at this point it's like arguing with a wall so I'm out. See you guys later. :)

Well, we basically agree then. As long as you take the complaints seriously and try to fix them instead of ignoring them because they are inevitable it should be fine.

3:50 PM

Oof, things are happening in the dead by daylight community it seems

user92578

4:41 PM

Aight, basic question, but: precompiled headers: what do I do when I need a type that's in a precompiled header in a header?

user92578

I.e. I put <vector> into my pch, and then I want struct Entity in Entity.h to have an std::vector<int> points

user92578

Does Entity.h include the pch? include <vector>?

Could you define "precompiled headers" ? What language is this?
As far as I'm aware, in C and C++, you don't really compile headers

Aaah

user92578

ahh I guess the headers implicitly get the data from the first source file that includes the precompiled header

In this case `Entity.h` could include either, and it would have the same effect.
If `Entity.h` is planning to use the `vector` object, it can include `<vector>` directly instead

user92578

4:44 PM

Main.cpp includes pch.h and then it includeds Entity.h, and then Entity.h can already access whatever I have precompiled

C++ handles recursive inclusion, and one of the reason of having header guards is that

There are a lot of ways to make this work, but best practise is, if a header needs a symbol from another header, it should include it itself

It eliminates headaches when you are refactoring

user92578

Right I see, looks like the thing is to keep headers as they are, but every source file gets the precompiled header as its first include

Headers are usually only for declaring symbols,
If you only plan to use a symbol (for example `vector`) in a source file instead, you can include it there instead

It's fine if both header and source file include the same other header

Behind the scenes, it will only be included once

user92578

I know how the include system works, but I have no experience setting up a precompiled header for a project

I'm not sure what you mean by "precompiled header", do you mean a header that someone else wrote?

4:57 PM

Me neither, but putting the precompiled header as the first include in every source file seems the be a rule to make them work.

headers are not meant to be compiled as far as I'm aware

See this or that.

user92578

You've never heard of precompiled headers?

I'm more of a C person, and so far never had to use them, but I see their point now

We need to use arrows more. I think everyone is confused who everyone is responding to.

user92578

5:01 PM

ohh now I see

user92578

you were replying to me nwp, I was replying to you & TomTsagk but now actually just TomTsagk

Yeah. I'm too slow at typing.

Any advice I've given until now was mostly referring to normal headers. I'm not sure if anything changes on precompiled headers :)

Things change. The idea is to compile the header and save the compiler state, then continue reading different source files, skipping having to re-read the same headers over and over. This can only work if the precompiled header is the first one because macros exist.

And none of that is in the standard, it's an extra feature compiler writers implemented in slightly different ways.

user92578

Do you happen to know of any effort to standardize this?

5:07 PM

“The wonderful thing about standards is that there are so many of them to choose from.” Grace Murray Hopper

I don't, besides modules, which sort of tackle the same problem.

user image

Relevant

user92578

Well AFAIK if we don't count versioning there's only one C++ standard

I would assume that whatever VS & GCC/G++ do are probably the de facto standards, but I'm not active w/ those languages, so that just my guess.

1 hour later…

6:12 PM

@Tyyppi_77 Yeah; I guess there are a lot of things that are complicated to implement? I don't know.

user92578

This might be 100% wrong but I feel like MSVC had really solid C++17 support pretty early

user92578

So a bit surprising about the state of 20

@TomTsagk that XKCD really made an impression on me. I had not thought of it that way before.

not this updated version

but the original

@Tyyppi_77 You put your precompile header in each of your .cpp file, being the first include. If you work with VS, and configure your solution to use precompiled headers, the compilation will fail if the cpp file is not set up as such. Then you just use the <vector>.

@Tyyppi_77 Maybe because c++17 was more of a fix over c++14? It seems it's a pattern they will follow.

user92578

But the big question was about includes in headers, which according to Google the solution is to not include the precompiled header in headers

user92578

6:19 PM

I feel like 14 was a fix over 11, but 17 was more of its own thing

user92578

Sure, mostly contexpr developments, but also filesystem, variant, optional etc, big stdlib additions that is

@Tyyppi_77 The Google is correct, the pch.h is only included in the source.

@Tyyppi_77 Right, that's probably that.

user92578

Apparently concepts is implemented in VS2019, I'm sort of tempted to give those a go... but I feel like it might be better to wait

I haven't had a chance to read about those yet. The list here is cool for the "features" but it's not super useful for direct information for what the thing is about. Links link to wording papers and such, and not to what are those for.

user92578

We need better patch notes :D

6:24 PM

:P

Sometimes I find that I don't have a purpose for the new thing. So I can't really go ahead and use it.

user92578

Do you have an example?

I think "modules" would be an example.

Most of our code is right into our app, and the code that is in a library will not be touched because we don't want to make other apps fail.

user92578

hmm true yeah, all though I haven't looked into those at all... I think those might make integrating 3rd party stuff easier though

user92578

assuming said 3rd party stuff is shipped as a module

Yeah; I do see the use, but for us, in the short term, there are none.

user92578

6:33 PM

We're at 11 at work for platform reasons, so all newer thigs are just home stuff for me

user92578

Plus we use 0 stdlib anyways

oh :/

huh?

"not made here"?

user92578

Hard to say what the actual reason is, the codebase dates way back, all though I'd say there's very little original code left

user92578

Custom containers etc. allow for cool features though, like we have versions of all containers with a local stack allocated initial buffer

user92578

I.e. if I know I probably have just a few items, I can use a cache vector that allocates 10 elements on the stack, and after that switched to the heap if more is needed

6:38 PM

Couldn't you achieve this with a custom allocator, though?

user92578

Maybe? I've never fiddled with custom allocators

Me neither :)

I should. This could help solve some memory framgentation issues we have.. :/

@Tyyppi_77 Do you think the code you wrote for the army is now legacy code, given the programmers generations that came after you? :P

user92578

Oh I didn't actually really write a lot of code there, I refactored a few messy scripts that were written before me for the tank I was told to work on, and then I got stuck trying to 3D model (not my thing at all) for the rest of my service pretty much

user92578

I ended up doing a lot of IT supportish stuff to help with the simulator trainings, browsed quite a bit of reddit, played a lot of solitaire :)

user92578

Worst programming language I've ever had to work with, most useful looking documentation was in abbreviated Czech

user92578

6:50 PM

"looking" as in looked useful, no idea if it actually was due to the language barrier :)

That's odd, why would they choose that kind of language?

user92578

The simulator is built ontop of a weird version of ARMA between ARMA1 & ARMA2

user92578

There was English documentation too, but the crucial pieces seemed to always be in comments inside code snippets, in Czech

Ah, well it explains the Czech there

1 hour later…

8:23 PM

god damn Shog9 quit

911

Q: Thank you, Shog9

@Shog9, a longtime Community Manager here at Stack Exchange, has just Tweeted that he is no longer working at Stack Overflow: @shog9: Well... I suddenly find myself in need of work. If anyone's hiring: C++, JavaScript, C#, SQL + solid background in social software, community-driven developm...

discussion featured stack-exchange-team thanks community-managers

user92578

Did he quit or was he let go? From his twitter post I assumed it was the latter

it's not clear, and nobody is commenting directly on it

might have been let go. hard to tell

8:40 PM

-383

A: Firing Community Managers: Stack Exchange is not interested in cooperating with the community, is it?

You all should know that this is very hard for us. This is what I am able to share: As you have seen, today there were some changes on the Community Team. Out of respect for their privacy, we aren’t able to comment on any specific current or former employees. We’re seeking to align the company...

"official comment"

user92578

> These changes are a part of that process.

Yeah, and..

People don't "suddenly" start looking for a new job unless they were fired or something so bad happened that they had no choice but to immediately resign. Either way, it's bad. — tvanfosson yesterday

yeah

Is someone at SE addicted to drama?

I'm starting to wonder if SE will survive this mess

4

it's just getting worse by the week

8:51 PM

I'm sure it will. People visit, users ask, users answer, content is generated, ads are displayed. It's just that the quality might drop a bit due to a bit of lack of enthusiasm.

user92578

hmm for some reason VS Class View shows that fmt leaks everything into the global namespace, which isn't the case... I wonder if they have trouble parsing their namespace macros

user92578

9:06 PM

Also my main annoyance with ImGui is that some stuff is under ImGui and some is just prefixed ImGui or even just Im

9:18 PM

@Vaillancourt good points

you probably right

@Tyyppi_77 i dont use that feature.

user92578

Well it has now crashed my VS three times so neither do I anymore

:p

1 hour later…

10:53 PM

@Almo I forgot: there is also SO for Teams. Maybe they'll end up making their profit only there, and use SO/se.com sites as an advertisement for that platform. Who knows!

eesh

11:12 PM

@Vaillancourt ive heard vs19 is pretty bad optimization wise.

user92578

as in it generates badly optimized code?

11:52 PM

no

social.msdn.microsoft.com/Forums/en-US/…

« first day (3449 days earlier) ← previous day next day → last day (1561 days later) »

Transcript for

Jan '2014

Game Development

Game development and other polite discussion. Game development...

game-design game-industry rubber-ducks

join this room
about this room

00:00

06:00

12:00

18:00

all times are UTC

site design / logo © 2024 Stack Exchange Inc; legal

mobile