« first day (2786 days earlier)      last day (2238 days later) » 

user92578
6:38 AM
"already??"?
 
(Unreal has )already( implement dx12 raytracing )??
 
user92578
yeah but how is this already?
 
better? :P
They were faster than light
 
user92578
you don't know that
 
user92578
they have probably started years ago
 
6:45 AM
Raytracing didn't get announced too long ago
 
user92578
because big game engine companies need to wait for the public release of gpu tech
 
Oh yeah... that makes sense
 
user92578
yeah it makes sense to send their stuff out early for real use like this, so they can get feedback on the stuff they're working on
 
The results look incredible though
So..
How's GunHero's development?
 
user92578
nonexistent this week, I've been super busy with course work
 
user92578
6:49 AM
gonna try to get a more done today and tomorrow so that I can dev on saturday & sunday
 
Nice
What's the course about?
 
user92578
uuuh I've been doing generic programming assignments for a generic programming course and physics assignments which are about basic quantum physics, and for tomorrow I have a few database ER-diagrams that I have to draw
 
8:12 AM
@Tyyppi_77 boooriing :D
I mean, it sounds so
 
 
1 hour later…
9:19 AM
Still getting upvotes on my basically useless answer to the leveled playing field question
 
user92578
@GabrieleVierti kinda yeah, quantum physics are really cool though
 
9:37 AM
0
Q: How can I give to player points with ajax but secured?

BenHow can I give to player points with ajax but secured? [on hold] Ask Question I have a virtual world game that I developed with jQuery, AJAX and PHP. Virutal World like club penguin. Now, I want to add a game into the virtual world - like bubbles or whatever. The main problem its that the game in...

He just copied his own closed question and posted it again
There's the [on hold] tag in the body
 
9:57 AM
@Tyyppi_77 quantum computers O.O
Now that's really interesting
 
user92578
Just the basic stuff is complex enough so the lecturer just briefly mentioned those
 
user92578
Very math heavy stuff
 
10:12 AM
0
Q: Capture Video without Third Party SDK's

Manoj Balaji MI am trying to capture the game play and store it on local memory without using any third party sdk's. I tried This Video Capture Script but i didn't get the recorded file even i don't know whether it's recording or not. anybody please help me to complete the task. Thanks in Advance

 
Wrote a container type for similar uses to std::vector, on my laptop it is officially badass:
--------------------------------------------------------------------------
Benchmark                                   Time           CPU Iterations
--------------------------------------------------------------------------
BM_create_vector_reserve/1                 61 ns         61 ns  117894737
BM_create_vector_reserve/8                 60 ns         59 ns  124444444
BM_create_vector_reserve/64                60 ns         60 ns  109268293
going to be adding more cases as time goes by, but for now .... boooya xD
(usv is mine, vector is the MSVC std::vector)
 
user92578
Nice
 
it's still not optimal at the micro level right now, but architecturally it seems sound
need to add some more fleshed out benchmarks doing some number crunching using it or something to get a better idea on the impact for a real workload
it does no heap allocation though so I expect memory I/O bound applications will be very happy
 
Anyone know why I'm getting : "undeclared identifier '_MainTex_ST" on this line "o.uv = TRANSFORM_TEX(v.texcoord, _MainTex);"
 
did you forget to save the file before compiling ?
also TRANSFORM_TEX might modify the name _MainTex
 
10:27 AM
lol nope
I'm thinking the same thing that it modifies it
 
simplest way to find out is to look at the implementation of it
 
so I defined a float4 _MainTex_ST
now it's saying that it is redefined at line 165 and line 185
my shader isnt even that long :D
 
As far as I know, the vertex shader created by your surface shader, will try to scale and bias the uv coordinates for the entries in your Input struct that contain the string 'uv', using the TRANSFORM_TEX defined in UnityCG.cginc.

TRANSFORM_TEX requires the name of the texture, for example '_SomeTexture', so that it can find '_SomeTexture_ST', which is automatically filled by Unity with the scale and bias information for that texture from the material inspector.
also
 
ok now I'm confused I think I'm writing a additional vertex shader inside a surface shader that compiles it's own
 
nwp
@OMGtechy What? So you are essentially comparing your std::array against their std::vector. That's not a fair comparison. Compare your vector against std::array instead.
 
10:37 AM
you can specify the size at runtime of mine, so I felt std::vector was a fairer comparison
you can't do that with std::array
would be interesting to see how it compares to std::array anyway though :)
 
nwp
How can you specify the size at runtime and not use heap memory?
 
move the stack pointer
 
nwp
How do return your container from a function?
 
you do not - that is one of its limitations - this is for people who do not need such things
 
nwp
Or put a million elements into it? (stack only has ~1MB of space)
 
10:40 AM
std::vector is very general - this is an optimizations for a special case that std::vector cannot make because it's general
Closer to C VLAs than anything else
 
nwp
Yeah, VLAs would be a fair comparison.
 
it provides a vector-like interface, so you can push/emplace back, query the size, etc
I would have used C VLAs internally but it just doesn't work because of scoping
 
nwp
I'm not sure if you can use a custom allocator for std::vector that uses alloca instead of malloc.
 
AFAIK you cannot
which is why this was made :D
the memory returned by alloca is free'd at the end of the calling functions scope
so if it's wrapped in an allocator it'll get free'd when the allocator returns
if you inline that, it'll get free'd when the vector ctor returns (or whatever)
an alternative would be to allocate a buffer outside of the vector and tell it to use that like a heap block / pool allocator would
 
@Tyyppi_77 ooohh.....
 
nwp
10:45 AM
You would need something like a vector_view that adopts a given piece of memory.
 
yeah that sounds like a good idea
this type exists to deal with all of that for you anyway
you don't need to allocate the buffer beforehand, you should say how big and go
 
nwp
Does it play nice if you have 2 of your vectors in a function?
 
yes
 
nwp
I would have expected fragmentation issues.
 
It will adjust the stack pointer at the point of construction
there's a debug check for alignment too
 
nwp
10:48 AM
So you can't resize after construction?
 
No, it has a fixed max size once you construct it
 
nwp
Oh. I guess there is still some benefit because the max size doesn't need to be constexpr.
 
yeah, it's for those cases where you could have used a C VLA, but want to keep a variable amount of things
so rather than manually keeping track of how many are in there etc, you can just use this
 
nwp
Have you seen this? I remember it talking about using alloca allocators and a framework to allow all kinds of cool behavior.
 
I have not - did they make an alloca allocator ??
I wasn't sure if it was possible
 
nwp
10:56 AM
They made cascading allocators. Things like "Use that global buffer. If it runs out of space use alloca. If that runs out of space use malloc" and the user would not need to know any of that.
 
I might take a look at the alloca part then - curious as to how they do it - thanks :D
 
 
1 hour later…
12:08 PM
working on my dynamic snow youtu.be/QnxlJt4WtaQ
 
@UriPopov I like it
 
thanks man still a lot of work to do on this one. I just added normal maping and specular glitters
 
nwp
But the snow just disappears and doesn't condense! My immersion! Literally unplayable!!111
 
ah I'm working on that too my man @nwp
 
nwp
(the !!111 are an irony indicator in case you missed that)
 
12:25 PM
@UriPopov awesome ! :D
 
@nwp it all ok man it all ok
@OMGtechy thanks
 
user92578
12:37 PM
That's damn cool Uri!
 
Thanks Tyyppi
 
1:31 PM
hmmm the recording fugged up again for some reason
 
1:43 PM
@nwp You can reserve more :P (And sometimes you have to, when your software crashes with a stack overflow in your physics engine.)
@KevinvanderVelden Pinned :)
 
Oh hey, I could have done that too!
 
@KevinvanderVelden Yeah, pinning is fun!
You'll be able to un-pin once the sale is gone :)
 
And pin the next sale :D
 
The next relevant one =p
 
1:51 PM
Yeah!
 
Cause pinning the... what was it a while ago? The "romance" novel sale would be a bit off-topic
 
hehe it could be used a inspiration for visual novels I guess :P
 
 
1 hour later…
3:09 PM
Is there even a need to render the head? Haven't played with VR for a while, so I'm not up to date on what's done nowadays..
 
 
1 hour later…
4:24 PM
@AlexandreVaillancourt you need the head for the artsy use of mirrors :)
 
nwp
It's a vampire world. Problem solved. Please transfer the estimated amount of development time required for proper mirror head rendering times your hourly wage to my account.
 
I refuse to play any games in VR without the use of mirrors or reflective spoons.
 
@lambparade Is there really a spoon, though?
 
@AlexandreVaillancourt oh my god...my mind is blown.
 
5:10 PM
void
MyClass::doSomething()
{
  // Get the Manager
  Manager* mgr = mResources.GetManager();
I love those useful comments. They allow me to reduce the size of the file when I delete them!
 
nwp
@AlexandreVaillancourt That was so funny chat broke.
I like that you see the positive side of things.
 
@nwp Haha :P
Can't complain all the time, people would mute me :/
 
nwp
Come to the lounge. It embraces whining. Just don't bring up C++.
 
Hehe I wish I had time for more chat :/
 
6:04 PM
5
It took you 5 messages, to get from VR to matrix
 
Lol, am I always talking about The Matrix?
 
6:40 PM
@UriPopov niice!
 
 
1 hour later…
7:49 PM
Great, C# and Java got to a new level of copying eachother, where java implements features from C#
Why the **** do we need a var in java?
 
nothing wrong with copying good ideas
 
That's not the problem
 
var is easy to implement (it doesn't require any changes in the compiler's data structures) and basically harmless (it doesn't propapage outside local scope)
 
Var gets abused a lot in C# (at least based on my experience) and it makes the code terribly hard to read
 
if you find it hard to read, you can choose not to use it
 
7:56 PM
Yes, but a usual workplace has teams
 
well your team has to agree on coding guidelines
in my language var is mandatory, so obviously I have a biased opinion :)
 
@AlexandreVaillancourt You as in plural you
 
Man it's so difficult to use const-correctness consistently in c++
@Jimmy uuuuuuh, i never liked "var"s or "let"s
 
I'm pretty sure you like it based on what languages you're used to writing
 
My very first language was python
Then Java, C#, and C++
 
8:09 PM
so which is your most commonly-used language? I imagine that influences your preference?
 
Sooo yeah, i guess so :P
C++
 
Python doesn't even have declarations, which is one step beyond type-inferred declarations
it's one step more than I like, since Python's choice forced them into weird ways to handle closured-over variables and globals
 
Yeah
Using "var" or "let" in C++ is just like using "auto"
Useful, but if abused it makes the code unreadable
 
Reminds me of this Haskell programmer I talked to who took offense at the phrase "Python is executable pseudocode", because his pseudocode started off with type declarations
 
@Jimmy :D
can Assimp be built statically?
it uses a multithreaded dll as runtime library
 
8:27 PM
@GabrieleVierti Do you know libGdx?
 
which is quite annoying, since you need to copy and paste that .dll wherever the executable is
@Curio mmmm nope
 
I was CURIOus...
 
oh god...
@Curio did you call yourself like that just to make this joke? :D
 
Nope
 
aight
makes sense
@Jimmy
Haskell is a standardized, general-purpose purely functional programming language, with non-strict semantics and strong static typing. It is named after logician Haskell Curry. The latest standard of Haskell is Haskell 2010. As of May 2016, a group is working on the next version, Haskell 2020. Haskell features a type system with type inference and lazy evaluation. Type classes first appeared in the Haskell programming language. Its main implementation is the Glasgow Haskell Compiler. Haskell is based on the semantics, but not the syntax, of the language Miranda, which served to focus the efforts...
that's so confusing
module Main where

main :: IO ()
main = putStrLn "Hello, World!"
what is this
 
8:32 PM
Haskell is a language where all Input/Output is tracked by the type system
which is why the Hello world is not immediately obvious
 
Yeah, it's a functional language. Things with side-effects like outputting to the screen are kinda weird
 
to be fair, there's a difference between "functional" and "pure"
in Ocaml, helloworld is print_string "Hello World"
 
Ocaml ?
 
a differen't ML-family functional language
more object-oriented functional flavor
 
oh okay
I've never used any of these languages..
@Jimmy weren't you developing a programming language?
 
8:37 PM
I am, although not much progress this month
 
user4704
@GabrieleVierti get a real deployment process
 
user4704
generally you want the runtime to be a DLL
 
assimp does have some kind of BUILD_STATIC_LIB flag in its cmake file
but I've spent like 30 seconds grepping for that on github and have no idea what it does and whether it actually works
 
@JoshPetrie so it wouldn't make sense to build it statically?
@Jimmy i haven't found anything about that..
 
"Update CMake environment to accept BUILD_STATIC_LIB option to allow building of static library for platforms that require it (such as iOS)"
oh wait nevermind
it was removed
 
8:42 PM
@Jimmy yeah ahahahaha that's what i was writing :D
so weird
if i build assimp with cmake-gui i get an the debug/release.... versions, only for 64bit windows
if i build it using cmake.exe, i get debug,release... versions for x86 windows
 
user4704
@GabrieleVierti You can build assimp statically
 
user4704
But static-linking the CRT is a different matter
 
user4704
You need to pass the appropriate -DWHATEVER switches to cmake. the generate you use also impacts what you will get
 
user4704
e.g., the makefile generate cannot produce "multi config" projects like the VS or Xcode ones can
 
user92578
8:54 PM
@GabrieleVierti how much are you moving stuff around?
 
@Tyyppi_77 just a dll file
i know
but it's still annoying
 
user4704
"Moving" where? And when?
 
user4704
Your build should put it in the right place for your development iterations.
 
user4704
Your installer should put it in the right place (or run the right redistributable; sometimes you cannot legally distribute the runtime DLLs) for the user.
 
user4704
If either of those are not true, that's a bug on your end.
 
8:58 PM
@JoshPetrie assimpblablabla.dll
 
user4704
I don't see the crt there
 
user4704
that's assimp's DLL.
 
user92578
But what has to be moved? Why cant stuff just stay where they are?
 
user4704
that's not the CRT
 
8:59 PM
that's what i don't want to move arounf
 
user92578
But what has stuff to be moved? Why cant stuff just stay where they are?
 
user4704
So why are you? That looks like the right spot.
 
user4704
Next to the .exe
 
@Tyyppi_77 "Core.exe" needs the .dll file
 
user92578
Yes and it has the .dll file right there
 
9:00 PM
yes
 
user4704
Is that after you've moved it?
 
user4704
Why isn't your build process configured to deploy/install the DLL there then?
 
user92578
Why are you building assimp so often that this is an issue?
 
because i have no clue on how to do it
@Tyyppi_77 because I've always build it dinamically, not statically
 
user92578
9:01 PM
Or are we talking about release packages here?
 
user4704
Are you using ExternalProject_Add to fetch assimp?
 
@JoshPetrie uuuuuhm... no
 
user92578
So everytime you build your game, you also build assimp?
 
@Tyyppi_77 no
 
user4704
Or are you doing add_subdirectory to include assimp's source?
 
user4704
9:02 PM
Or something else?
 
user92578
I just dont understand what happens to the DLL
 
user4704
I have a feeling I do and it's going to be bad :P
 
@JoshPetrie no, I'm not doing any of that.... i'm starting to worry
 
user4704
I have a feeling that right now he's running cmake on the assimp directory manually and manually building it "whenever needed"
 
user4704
and then copying it
 
user4704
9:03 PM
rather than having HIS CMakeLists.txt handle the dependency with add_subdirectory, ExternalProject_Add, etc
 
Yuuuup
 
user4704
that's why it has to be manually copied, because his build process neither builds it nor understands it as a target and cannot do it for him
 
user4704
Yeah don't do that :P
 
user4704
You use CMake for your own project, as well?
 
user4704
Or do you just use some IDE's project files directly?
 
user92578
9:04 PM
But like shouldn't that be okay? Build assimp once -> move to output dir -> keep on working using the built DLL
 
user92578
My DLLs have happily lived in the Debug folder for years
2
 
user4704
@Tyyppi_77 Until assimp gets updated, yeah. And also it's not a one-step build, which is less clean than it could be (but servicible)
 
user92578
Some of them built manually, it's not a big deal to update one every now and then
 
@JoshPetrie mmmmmm no....
 
user4704
Whereas, for example, I have this in my cmakelists:
 
9:06 PM
Not that familiar with Cmake as you can see
 
user4704
fetch_dependency(FMT STATIC
  ROOT Dependencies
  URL  github.com/fmtlib/fmt.git
  HASH f61e71ccb9ab253f6d76096b2d958caf38fcccaa
  INCLUDES Deploy/include
  BINARIES
    Deploy/lib/libfmt.a FMT
)
 
user92578
Well that's sorta cool, CMake does have its benefits yeah
 
user4704
And all I have to do is change the tag to cause the next build to fetch, build and deploy the latest version of libfmt into my project. It also automatically happens on the first "clean" build after checkout
 
user4704
@GabrieleVierti What do you use then? VS?
 
user92578
On the other hand I'd like to decide when I update dependency X
 
user92578
9:07 PM
Oh nvm I misunderstood
 
user4704
@Tyyppi_77 I do; don't change the tag, dependency doesn't update.
 
@JoshPetrie i guess :D
 
user4704
This way I don't have to actually check the binaries in to source control, which is nice.
 
user4704
OTOH the very first clean build of a fresh clone can take a while.
 
user4704
Since it has to fetch and build all dependencies.
 
user4704
9:08 PM
bullet is particularly nasty to wait for
 
user4704
@GabrieleVierti You can have your project file run a prebuild script to copy files around, for example
 
user4704
probably simpler than learning how to use cmake for your own stuff
 
user4704
that requires a bit of a commitment
 
user92578
This is starting to make sense, I'm also realizing that this cool stuff is totally unnecessary for me
 
user4704
Yeah, it's not something you really need to do if you don't distribute your stuff around.
 
9:11 PM
@JoshPetrie make the project run a script?
 
user4704
Yes.
 
user4704
you can tell it to execute some batch file
 
user4704
and in that batch file, put the commands to run cmake, copy the outputs where you need them, etc.
 
user4704
or just deal with moving the file now and then
 
Are we talking about the V
 
user4704
9:12 PM
shrug
 
VS project?
 
user4704
yes
 
user92578
My builds run Python scripts, it's handy to be able to run stuff before builds
 
user4704
in the project properties dialog someplace is a section for prebuild and postbuild actions
 
user4704
(you could also do it directly in the msbuild of the project itself but that's also complicated)
 
user92578
9:13 PM
Oh actually I think I use the make stuff in VS for that
 
user4704
yeah that's another option
 
user92578
That like checks for changes first
 
@JoshPetrie ooh niice :D
How are you guys doing?
Guess everyone is dead :P
 
Curio is alive
 
@Curio Rejoice! \o/
 
9:25 PM
:D
 
Lol :D
 

« first day (2786 days earlier)      last day (2238 days later) »