What are you working on? v67 - March 2017 - Developers

Yeah, I had a go at this in some primitive form. Before lcef, I was trying to figure out how to write a CSS 2.1-compliant box model based GUI module to replace what exists in Grid 7, which is fully imperative, and use XML and CSS or some shit like that and make it declarative instead. It would have been a compromise in speed (lower overhead, not all of Chromium, holy shit) and ease of programming experience. But it turns out I hate writing software so fuck that.

i made a discord bot to show my faceit matches, but i hate it. i keep losing and it keeps getting recorded :< https://i.imgur.com/IQoaMP1.png

I've been doing this as well, and it's certainly a lot of work, but at least I don't have a few million lines of code in dependencies I'm using XML and CSS, and I have the most basic CSS properties supported already (display, border, margin, padding, background, font-family, font-size, color, overflow, etc.). Layout is mostly working, with the exception of some corner cases for absolutely positioned elements. Of the input elements, I currently support text input (which was another can of worms because Unicode is complicated) and checkboxes. The code needs some optimization and clean up, but I haven't had time for that yet. I gotta implement sliders and comboboxes at some point still I've been thinking of separating this from the engine into its own library, but I'm not sure if it'd be worth the effort

In competitive highschool and college debate, cases are big gigantic files full of evidence for your side (Aff or Neg, Pro or Con), and a tool was released several years ago and is still updated in order to format debate cases. This tool, called "Verbatim", has been pretty much the standard for competitive debate for several years. Essentially it's just a MS Word template so that different sections of a case, headings, citations, etc. can be formatted in a way that's common and recognizeable to all debaters. I want to go about serializing debate cases, and Verbatim has made that really easy for me. Since these are almost always in .docx format, I simply need to parse out the XML for a document. For example: https://files.facepunch.com/forum/upload/132997/d9426970-95c3-47ba-8353-14bf76c7f5e1/soffice.bin_2018-04-07_10-36-57.png This is one "contention" in a debate case I found online. Most of them look like this. Those bolded and underlined words are the emphasis added by the person who wrote the case, but everything in between that and the citation is someone else's article (Wilson, 15). Anyways, since these are all made using Verbatim, I can find common headings and tags in the XML: <w:p w14:paraId="7944F8CC" w14:textId="77777777" w:rsidR="00AD2AFA" w:rsidRPr="00DA731F" w:rsidRDefault="00AD2AFA" w:rsidP="00AD2AFA"> <w:pPr>  <w:pStyle w:val="Heading2" /> <w:rPr> <w:rFonts w:ascii="Calibri" w:hAnsi="Calibri" /> </w:rPr> </w:pPr> <w:r w:rsidRPr="00DA731F"> <w:rPr> <w:rFonts w:ascii="Calibri" w:hAnsi="Calibri" /> </w:rPr> <w:lastRenderedPageBreak /> <w:t>Part 3 is Offense</w:t> </w:r> </w:p> So most of the headings and bodies, etc. can be found associated with "Heading1", "Heading2", "Heading 3", etc., uniformly followed by a citation with the author's last name and date of publish in bold. Personally, I think there's a lot of intellectual dishonesty in competitive debate. The amount of space in between emphasis added in the above selection goes beyond the point of abridging the article for the sake of brevity and crosses into truncating entire paragraphs and sentence fragments to manipulate the Author's intended meaning to support a certain side. You could argue that that's what debate is about (manipulating the facts to support one side), but I think there's a LOT wrong with competitive debate: https://www.youtube.com/watch?v=0FPsEwWT6K0&t=10s

https://i.imgur.com/KeM6kKC.png

why in the pooperdooper are they hashing rendertargets

So I haven't posted here in eons by now, so I was busy making games and inserting my face into linear algebra, in other words I also recently put dear imgui into Godot 3.0.2 through it's GDNative thing (where you can load native code as shared libs without recompiling the engine), so I'm using imgui (C++ library) in Godot (C++ engine) through a C-binding to dear imgui as a shared library (cimgui) through derelict-imgui (the D binding) which I'm loading in my code I'm writing in D towards towards Godot with godot-d! But hey, now I can debug my gameplay code with imgui just as I wanted to.. and all this because I wanted to put debug ui on screen quickly (Godot's UI is nice but it's a lot of work just for debug stuff). (don't mind the rekt vertice at the end of every set of text, still figuring out why that happens) Still, I'm happy it was totally possible! https://files.facepunch.com/forum/upload/123559/f9579c74-e32a-4625-9e2a-df483499f963/datimgui.png

another questionable love decision

my free google drive 15gb of space is almost running out from one document containing questionable decisions love devs make.

Just finished this little piece of open source code: https://github.com/viliwonka/WeightedRandomSelector/blob/master/README.md Tried to make it elegant but in the end just rushed it so I don't spend too much time on it. Opinions?

i added some chatting to this game that i've been working on https://my.mixtape.moe/moxqad.mp4

Learning about Neural Networks/Deep Learning. Have a potentially billion dollar idea involving Rust, Cryptocurrency, Blockchain, and AI. Just kidding But I am learning about Neural Networks for a project that could be worth a little bit of money. My new company won't let me earn money from my own code while I work for them :/

That background noise

That's pretty bullshit if you do it in your own time

I've been working on an online EBX (Enhanced Binary XML) viewer. We're creating a Battlefield 3 mod, and I found it tedious to search through text files to find the references and whatever. [t]https://files.facepunch.com/forum/upload/132348/1a66308f-2cba-4c0f-81fd-51f87a4acb97/image.png[/t] It even supports blueprints. [t]https://files.facepunch.com/forum/upload/132348/11f1e236-bf9a-4659-87d3-ba1ee43450b3/image.png[/t]

title of your sex tape

They consider it to be consulting and it falls under my non-compete clause

Tracking down a bug at work that baffles me. I've found the revision at which it initially occurred(some 400 revisions ago) It occurs in latest the same way it does in that revision. Turning off pre-loading of UI objects only gets rid of the crash in that old revision, it doesn't change anything in latest. Running the stomp allocator on the old revision causes the crash to not happen, still occurs on latest. The crash itself is also a complete mystery, I get a nullptr in the PhysX library while simulating a kinematic body. However, it doesn't seem to be a memory corruption issue because all the memory around the nullptr is valid. But it's almost as if it can't be anything else. What are the chances of some allocation logic breaking only a single 64-bit pointer, leaving all surrounding data correct?

Is that normal practice? fuck that

What a load of shit

From what I know that's pretty normal in the industry

What is the most common method of handling latency in multiplayer games? For example if a player with 300ms of latency changes their client state to move forward, how should the server deal with the differences in time, so the player won't appear to rubber band around the world?

I just read my employment agreement in more detail. It actually says that I can "moonlight" so long as 1.) I'm not competing with my employer 2.) I'm not skipping work to work for someone else 3.) I'm missing deadlines because of other engagements 4.) I get permission from my boss first

FUCK. This class of bug terrifies me. It's the only time in software engineering that I feel like an idiot. You're working with a custom allocator(?)... which is terrifying... are you also working with threaded code? If the crash is entirely unrelated, it sounds like you could have some sort of access violation from threaded code. It sounds like the stomp allocator resolves one issue, but you actually have two or more (compounded[?]) problems.

if you suspect you might have a memory corruption issue, you should run your app with page heap in application verifier enabled i think it comes with the windows 10 sdk nowadays Windows 10 SDK although if you have a custom allocator that uses its own heap or something i don't think it'll work in that case

It's inside of Unreal Engine so yes, all of the things you mention we can tick off on. The physx simulation is a task on a thread, but the task itself isn't multithreaded. I think the stomp allocator just makes it so that the problem lies somewhere else instead, where it isn't as much of a disaster. Or Physx has its own giant memory block allocated so that the memory corruption inside of the physx memory can't be detected unless PhysX has their own. I'm not familiar with PhysX at all though so I have no idea if that is even remotely possible.

I really love what you've been doing with Source, Sam. It definitely inspires me to play around with the SDK again. Thought you should know since very few good Source devs are still around.

That gif makes my eyes hurt because it's such an ungodly mess. Good job.

I can imagine. Plus, Mike died, so there's that.

he died? i'd thought he just left valve