[QUOTE=paindoc;52837520]i pity the guy here who did the actual work in moving us from SVN to git, but somehow he seemed to pull it off
I like sourcetree, but I've started using gitkraken. at first I didn't like how it really pushed the "modern" aesthetic and it seemed to almost obfuscate functionality, but gitkraken handles stashes better, auto-fetches from the remote much more frequently, and has handled larger repositories better. I've also found its conflict-resolution tools to be much better than sourcetree's.
Also, sourcetree has been super buggy and crashy for me as of late.
[editline]30th October 2017[/editline]
also, things like this are why I'm pretty sure I'll never truly know C++ in full. I constantly discover new things, like this.[/QUOTE]
It's because they keep adding new shit :v:
[QUOTE=Leystryku;52837559]It's because they keep adding new shit :v:[/QUOTE]
yeah the committee slept for a fucking decade then finally decided it was time to get woke
the absurd difference in featureset has become even more apparent due to the work I am doing now, since this codebase was written up to two decades ago (and the newest stuff is still ~15 years old). like how there's no need for a custom std::string implementation, new/delete can be safely banished, std::array can replace c-style arrays, constexpr replaces #define, nullptr replaces NULL, std::move adds several effectively free perf improvements, etc and etc.
as an example of how fucking C++ keeps goddamn surprising me - there are functions for calculating [URL="http://en.cppreference.com/w/cpp/numeric/special_math/legendre"]legendre polynomials[/URL], something I was currently stuck on in my lunar geopotential model implementation. I had glanced at the special math stuff in the past, but had never noticed the legendre polynomials before trying to implement my own. goddamnit :V
[editline]edited[/editline]
actually, i need the associated legendre polynomial. [URL="http://en.cppreference.com/w/cpp/numeric/special_math/assoc_legendre"]but it supports that too[/URL]. ffs.
I used to use sourcetree but it kept breaking every time it updated and started running really poorly. Plus the old interface it had was much better IMO
I've switched over to the official github app and had 0 issues since and it worked perfectly straight from the get go.
[QUOTE=paindoc;52837606]yeah the committee slept for a fucking decade then finally decided it was time to get woke
the absurd difference in featureset has become even more apparent due to the work I am doing now, since this codebase was written up to two decades ago (and the newest stuff is still ~15 years old). like how there's no need for a custom std::string implementation, new/delete can be safely banished, std::array can replace c-style arrays, constexpr replaces #define, nullptr replaces NULL, std::move adds several effectively free perf improvements, etc and etc.
as an example of how fucking C++ keeps goddamn surprising me - there are functions for calculating [URL="http://en.cppreference.com/w/cpp/numeric/special_math/legendre"]legendre polynomials[/URL], something I was currently stuck on in my lunar geopotential model implementation. I had glanced at the special math stuff in the past, but had never noticed the legendre polynomials before trying to implement my own. goddamnit :V
[editline]edited[/editline]
actually, i need the associated legendre polynomial. [URL="http://en.cppreference.com/w/cpp/numeric/special_math/assoc_legendre"]but it supports that too[/URL]. ffs.[/QUOTE]
What the fuck, they put legendre polynomials in the standard library? I've been struggling with that in the past.
[QUOTE=DrDevil;52837968]What the fuck, they put legendre polynomials in the standard library? I've been struggling with that in the past.[/QUOTE]
msvc doesn't support it yet, but it should be in soon
Picked up trying to learn UE4 again finally. Decided that maybe I should give blueprints a stab even though I already can program in c++ because they seem pretty cool. First thing I wanted to do was create a simple menu and do some options menu key-rebinding shit. Nah apparently that's not something UE4 supports in blueprints? That's a pretty basic fucking feature, am I missing something here? I can't click a menu button, listen for any key input and then rebind a set actioninput with said key? Am I missing something here? Because I feel like I must be, but literally zero UE4 documentation that I've found has mentioned this anywhere. The most I've gotten was some guy's plugins from outdated versions that hacked their way around it by just making some custom blueprint stuff with c++. I know I can do it easily in c++, I'm just really amazed that there's apparently no native support for it.
I feel like I shouldn't feel like I'm drowning just trying to learn the basics of an engine when I've done all that shit and more my self before. Game development in a streamlined environment specifically tailored for it shouldn't be this much of a headache just to get a simple menu working. Who the fuck at Epic said, "yeah so this big feature that we advertise as a complete thing that you can use without knowing any code at all? Make it so we don't have any blueprint functions to handle remapping this event action input system we developed specifically for it. Who needs basic I/O remapping anyway?"
Am I missing something really obvious here or what?
I haven't posted in one of these threads in years, but I finally have something kind of worth talking about. I'm working on finishing up my compiler for a stack machine simulator, the architecture for which is based on the [URL="https://en.wikipedia.org/wiki/Burroughs_large_systems"]Burroughs B6700[/URL]. I have about three days to do code generation, which I missed the lectures for so I'm kinda doing this by feel. I'll probably need to drop some major features like arrays in favour of just getting the majority of the functionality working. Please send coffee and prayers.
It's super cool to have gotten this far though, I'm at the point where I can feed in any correct program and have it parse it ready for generating instructions. It's a little light on the semantic analysis but that only matters if you make mistakes :v:.
For example, a super simple program:
[CODE]
CD testprogram
main
begin
Out << 5 << Line;
end
CD testprogram
[/CODE]
Is converted into this tree:
[IMG]https://i.imgur.com/aUu8YUI.png[/IMG]
And becomes these instructions, to output the number 5 to stdout.
[CODE]
1
41
5
64
67
0
0
0
0
0
0
0
[/CODE]
I just wrote this hack here because the properties in Kotlin [I]object[/I]s aren't enumerable (in JS): [code]fun enumerablify(obj: Any, depth: Int): Any {
val newObj = Object.create(when (depth) {
0 -> Object.getPrototypeOf(obj)
else -> enumerablify(Object.getPrototypeOf(obj), depth - 1)
})
for (propertyName in Object.getOwnPropertyNames(obj) logged "Properties") {
val descriptor = Object.getOwnPropertyDescriptor(obj, propertyName) logged "Descriptor"
descriptor.enumerable = true
Object.defineProperty(newObj, propertyName, descriptor)
}
return newObj
}[/code]
It's used like this: [code] val tempDefinition = object : ComponentDefinition {
override val schema: Any
get() {
log("Schema gotten.")
return object {
val msh = object {
val type = "model"
}
}
}
override val init: (ComponentPrototype.() -> Unit)?
get() = { log("init") }
override val update: (ComponentPrototype.() -> Unit)?
get() = { log("update") }
override val tick: (ComponentPrototype.() -> Unit)?
get() = { log("tick") }
override val remove: (ComponentPrototype.() -> Unit)?
get() = { log("remove") }
override val pause: (ComponentPrototype.() -> Unit)?
get() = { log("pause") }
override val play: (ComponentPrototype.() -> Unit)?
get() = { log("play") }
}
val definition = enumerablify(tempDefinition, 1) as ComponentDefinition[/code]
Note that I had to use a [I]depth[/I] of [I]1[/I], since the properties belong to the prototype.
I didn't make the function generic because this is a hack and possibly breaks some things on the Kotlin side. I only need to pass this [URL="https://github.com/aframevr/aframe/blob/188f653ea33a327afdd979fea4fc7c30f7d25723/src/core/component.js#L458"]into JS[/URL], so there's no issue here in that regard.
[editline]edit[/editline] Don't mind the horrible syntax of that [I]object[/I]. I'm using [I]external[/I] definitions instead of a proper wrapper right now, so some things are pretty messy.
[QUOTE=F.X Clampazzo;52839187]Picked up trying to learn UE4 again finally. Decided that maybe I should give blueprints a stab even though I already can program in c++ because they seem pretty cool. First thing I wanted to do was create a simple menu and do some options menu key-rebinding shit. Nah apparently that's not something UE4 supports in blueprints? That's a pretty basic fucking feature, am I missing something here? I can't click a menu button, listen for any key input and then rebind a set actioninput with said key? Am I missing something here? Because I feel like I must be, but literally zero UE4 documentation that I've found has mentioned this anywhere. The most I've gotten was some guy's plugins from outdated versions that hacked their way around it by just making some custom blueprint stuff with c++. I know I can do it easily in c++, I'm just really amazed that there's apparently no native support for it.
I feel like I shouldn't feel like I'm drowning just trying to learn the basics of an engine when I've done all that shit and more my self before. Game development in a streamlined environment specifically tailored for it shouldn't be this much of a headache just to get a simple menu working. Who the fuck at Epic said, "yeah so this big feature that we advertise as a complete thing that you can use without knowing any code at all? Make it so we don't have any blueprint functions to handle remapping this event action input system we developed specifically for it. Who needs basic I/O remapping anyway?"
Am I missing something really obvious here or what?[/QUOTE]
The UI portion of the Unreal Engine is a total nightmare, UMG is a hack built upon the barely useable slate, so that certainly doesn't help.
But yeah, in my experience, Blueprints are useless if you can program in C++. The only benefit blueprints have is that less experienced programmers or designers can use C++ code to make menus or small game mechanics, deal with SFX and VFX etc etc
I was under the impression blueprints were pretty much Unreal's solution to swapping out constants for testing without having to recompile all the C++ code. Whereas in something like Unity you have the editor variables and C#'s fast compile and deploy time.
Also a gimmick to draw in the indie crowd.
I feel like blueprints were made to quickly edit variables and hook up very basic things, kind of like a more advanced prefab in Unity. Then you have people making whole games using them and seeing it as an accomplishment. No, all you've done is made a spaghetti mess that has made it harder in the long run.
The main argument I get for blueprints is that it allows artists to code, if artists want to code then why not learn to code or leave coding to the coders?
I really like blueprints, they allow to quickly cobble stuff together and rapidly test ideas.
[QUOTE=DrDevil;52840656]I really like blueprints, they allow to quickly cobble stuff together and rapidly test ideas.[/QUOTE]
Honestly, I find visual scripting to slow me down, having to create the boxes and wire them all together, for example, a single line of code could be 5-20 components in blueprint. That's just my opinion.
[t]http://powback.com/u/59f8a9f2953d0.png[/t]
The Frostbite Map Editor now fully loads maps. Only textures are missing. After that, we'll do Entities like lights and what not.
[QUOTE=0V3RR1D3;52840849]Honestly, I find visual scripting to slow me down, having to create the boxes and wire them all together, for example, a single line of code could be 5-20 components in blueprint. That's just my opinion.[/QUOTE]
Yeah, that's true, it happens me the same with playmaker although I tried too hard to give it a value that it costed me, but after a while I just found that even for simple things, it just took more time than it should
[QUOTE=0V3RR1D3;52840849]Honestly, I find visual scripting to slow me down, having to create the boxes and wire them all together, for example, a single line of code could be 5-20 components in blueprint. That's just my opinion.[/QUOTE]
Well, using a visual language requires a different thinking. For example I find it much easier to divide my "code" into functional blocks by just visually bunching them together. It also make experimenting easier, as you don't have to copy paste lines of code around, but can just reroute the blocks.
It's a matter of taste I suppose. Just don't dismiss it as a toy, because it is very powerful.
[editline]31st October 2017[/editline]
It also is better at visualizing execution paths, so complex game logic and decision paths could become less convoluted to work with in form of a graph.
This model I'm working on is kind of the perfect storm of difficulty:
- It tends to create numbers that are both extremely large and extremely small, and then performs operations between these numbers hugely separated in range (rip accuracy)
- Getting around these errors usually just requires fuckloads of recursion
- to make threading tougher, each of the three geopotential calculations (split among three directions) has to share access to a single constants table
This is almost as awful as making that fucking random walk pattern for 3D print infill. I mean, I like a tough challenge but I'm on a serious timeline here and no one else has tried to use a similar spherical harmonic geopotential model for the moon, yet. so its tough to know if I'm even doing this right at all. So long as I keep the order and degree low, results seem fine. Once the degree gets higher, the legendre polynomial begins to head to tons of overflow and underflow, and this number then gets multiplied by really small trigonometric and constant values. So the accuracy just goes to complete shit, fast.
I get accelerations with numbers like "3.09x10^131", which just doesn't seem [I]quite[/I] right
Hey guys, long time lurker here. I'm a big fan of WAYWO and I've been visiting for a couple years; all of your work is truly inspiring (and makes me feel like a lazy cunt). With that said, I just "released" a small game I made in 30 days! It's a randomly generated horror survival game called Midnight. The goal is to get as deep as you can, I guess. I hope to improve it more over the coming months. I would greatly appreciate any and all feedback, including bug reports, performance issues, etc. Here's a link for those of you who are interested in trying it out: [url=https://www.dropbox.com/s/mew4utostsvfgy6/Midnight.zip?dl=0]Midnight[/url]
Does anyone know if it's possible to completely redesign a GUI in Linux? My googling produces irrelevant results. By redesign, I mean turn regular GNOME into something that resembles a menu from Ace Combat.
[QUOTE=Karmah;52837801]I used to use sourcetree but it kept breaking every time it updated and started running really poorly. Plus the old interface it had was much better IMO
I've switched over to the official github app and had 0 issues since and it worked perfectly straight from the get go.[/QUOTE]
FWIW: Sourcetree's team is super small -- just 2 people for each OS.
Sometimes it can be difficult developing software that doesn't have bugs, or adds performance issues, for such a large number of users when you have so many features (some of which probably shouldn't exist anymore). You wouldn't believe the number of unique repos -- in size and history -- that are out there. Some features were also designed with smaller repos in-mind: opening multiple repos at once (kraken doesn't allow that), or jumping to a SHA in the log view (which may exist over 5,000 commits in the past...so you gotta load them all because the user will want to scroll in that history too!). Also we gotta support Hg, GVFS, LFS, and GitFlow as well, so that's another layer of complexity.
I guess what I'm trying to say is that computers are hard and sometimes we break things. But it's still free and not Electron, so that's kinda cool right?
Wellp, I have the most complex part of my menu handled now. Fucking only took a few hours to trudge my way through Epic's abysmal documentation on slate and shit to create a menu that can rebind actions. I don't even have anything else but that one menu. I'm eager to get actual creative stuff done, but file I/O for settings and saves comes next so I guess that'll have to wait.
[QUOTE=Foda;52842786]FWIW: Sourcetree's team is super small -- just 2 people for each OS.
Sometimes it can be difficult developing software that doesn't have bugs, or adds performance issues, for such a large number of users when you have so many features (some of which probably shouldn't exist anymore). You wouldn't believe the number of unique repos -- in size and history -- that are out there. Some features were also designed with smaller repos in-mind: opening multiple repos at once (kraken doesn't allow that), or jumping to a SHA in the log view (which may exist over 5,000 commits in the past...so you gotta load them all because the user will want to scroll in that history too!). Also we gotta support Hg, GVFS, LFS, and GitFlow as well, so that's another layer of complexity.
I guess what I'm trying to say is that computers are hard and sometimes we break things. But it's still free and not Electron, so that's kinda cool right?[/QUOTE]
No argument there. It was just a matter of convenience on my part.
I'm not a shining example of a proper user, but it seemed hell bent on installing itself into the User's temporary folder on a different harddrive then the one it was installed on every time it updated, and it performed a lot slower with its new UI then its older one on my machine.
The final nail in the coffin was it failing to authenticate my user account with bitbucket so I just said bitfuckit and switched to git 100%
[QUOTE=Adelle Zhu;52842516]Does anyone know if it's possible to completely redesign a GUI in Linux? My googling produces irrelevant results. By redesign, I mean turn regular GNOME into something that resembles a menu from Ace Combat.[/QUOTE]
Like [url=https://i.ytimg.com/vi/MkRC0lIUJmE/maxresdefault.jpg]this?[/url] You can't easily get Gnome to look like that, but you can use other apps like [url=https://github.com/DaveDavenport/rofi#application-launcher]rofi[/url], [url=https://github.com/brndnmtthws/conky]conky[/url], lemonbar, and others to get something close.
Check out [url=https://www.reddit.com/r/unixporn/]unixporn[/url] on reddit. Lots of people post configs, examples, and images of various desktop configurations. Most of them end up looking a bit like Ace Combat.
[QUOTE=Bael;52842347]Hey guys, long time lurker here. I'm a big fan of WAYWO and I've been visiting for a couple years; all of your work is truly inspiring (and makes me feel like a lazy cunt). With that said, I just "released" a small game I made in 30 days! It's a randomly generated horror survival game called Midnight. The goal is to get as deep as you can, I guess. I hope to improve it more over the coming months. I would greatly appreciate any and all feedback, including bug reports, performance issues, etc. Here's a link for those of you who are interested in trying it out: [url=https://www.dropbox.com/s/mew4utostsvfgy6/Midnight.zip?dl=0]Midnight[/url][/QUOTE]
Post a video or screenshots please, can't play it on linux :(
[editline]1st November 2017[/editline]
[media]https://www.youtube.com/watch?v=9wdQGoG3fYY[/media]
[url=https://www.shadertoy.com/view/llBczc]Calm Nebula[/url]
I've been browsing through shadertoy the other day and stumbled upon something incredible. By using the frame's previous content to construct the next frame's content, you can create absolutely unpredictable and chaotic behavior. Sometimes it happens that this behavior is stable on a macroscopic periodic scale, and thus produces beautiful patterns.
My shader is based on [url=https://www.shadertoy.com/view/XsSfWR]this[/url], which is just absolutely gorgeous. There is no input, the inherent instability of the system is enough to create a chaotic yet recognizable pattern.
Working with this type of shaders is a bit tedious, as the outcome is never predictable. If one value is set wrongly enough, the system is driven into instability and diverges into infinity (or converges towards zero which is equally uninteresting). I'd love to use them in live performances, but that can only realistically be done by finding a set of good values beforehand and only using those without too much deviation.
[QUOTE=Karmah;52843289]No argument there. It was just a matter of convenience on my part.
I'm not a shining example of a proper user, but it seemed hell bent on installing itself into the User's temporary folder on a different harddrive then the one it was installed on every time it updated, and it performed a lot slower with its new UI then its older one on my machine.
The final nail in the coffin was it failing to authenticate my user account with bitbucket so I just said bitfuckit and switched to git 100%[/QUOTE]
If you know git 100% then you have full control and it's fine.
You could also try the popular [url=http://gitextensions.github.io/]Git Extensions[/url]. We've all switched to it in our company except for our mac users.
nothing like 86mb text files to start your day
[editline]edited[/editline]
that being 86mb of text denoting a trajectory for a satellite i need to test my math against
[I]fuck[/I]
[QUOTE=paindoc;52844475]nothing like 86mb text files to start your day
[editline]edited[/editline]
that being 86mb of text denoting a trajectory for a satellite i need to test my math against
[I]fuck[/I][/QUOTE]
At a previous job I was regularily tasked with plotting 5gb large files containing data over a 3d surface with matplotlib. I worked on a window station, and they only had 32-bit notepad++ installed, so I couldn't even open the damn file without moving it to a linux workstation first to split it into two smaller pieces.
THAT was a pain.
[editline]1st November 2017[/editline]
Also couldn't just install my own software without having to track down the admins for hours of course, they sure had a talent for getting the fuck out of harm's way!
[QUOTE=DrDevil;52844555]At a previous job I was regularily tasked with plotting 5gb large files containing data over a 3d surface with matplotlib. I worked on a window station, and they only had 32-bit notepad++ installed, so I couldn't even open the damn file without moving it to a linux workstation first to split it into two smaller pieces.
THAT was a pain.
[editline]1st November 2017[/editline]
Also couldn't just install my own software without having to track down the admins for hours of course, they sure had a talent for getting the fuck out of harm's way![/QUOTE]
i'm glad my company is so small and informal, I wanted to use ubuntu on my work machine so i installed the software i needed to move data to the front of my disk, moved the data, created a new partition, installed ubuntu on said partition, and did all of this without asking anyone for permission lol
NASA has a really neat catalog of open-source software - [URL="http://gmatcentral.org/display/GW/GMAT+Wiki+Home"]GMAT might just save my ass[/URL], since it some pretty detailed mathematical models including spherical harmonic gravity. It is, of course, written by scientists (and this is painfully obvious) but I've seen worse and at least it has a CMakeLists file
[QUOTE=Karmah;52843289]No argument there. It was just a matter of convenience on my part.
I'm not a shining example of a proper user, but it seemed hell bent on installing itself into the User's temporary folder on a different harddrive then the one it was installed on every time it updated, and it performed a lot slower with its new UI then its older one on my machine.
The final nail in the coffin was it failing to authenticate my user account with bitbucket so I just said bitfuckit and switched to git 100%[/QUOTE]
hey -- no worries! we make mistakes and I don't take it personally when someone doesn't use ST for whatever reason :happy:
[QUOTE=paindoc;52844584]i'm glad my company is so small and informal, I wanted to use ubuntu on my work machine so i installed the software i needed to move data to the front of my disk, moved the data, created a new partition, installed ubuntu on said partition, and did all of this without asking anyone for permission lol
NASA has a really neat catalog of open-source software - [URL="http://gmatcentral.org/display/GW/GMAT+Wiki+Home"]GMAT might just save my ass[/URL], since it some pretty detailed mathematical models including spherical harmonic gravity. It is, of course, written by scientists (and this is painfully obvious) but I've seen worse and at least it has a CMakeLists file[/QUOTE]
I've tried to create my own orbital planning tool a few years back and used GMAT as a reference. After reading into the code I wanted to kill myself though... wxwidgets sucks.
Sorry, you need to Log In to post a reply to this thread.
What are you working on? v67 - March 2017
