Post AYpl1deWNuCDt86TvE by WAHa_06x36@mastodon.social
(DIR) More posts by WAHa_06x36@mastodon.social
(DIR) Post #AYpRvsNsnl3Ci8rMzA by mcc@mastodon.social
2023-08-17T03:19:22Z
0 likes, 0 repeats
ME: Hmm… a git repository from 2008 with C++ code and a CMake file. I will simply run CMake to compile it. There is no reason why C++ code from 2008 should not compile fine now.CMake: C:\Users\Andi\work\g\other\tangere\Common/FileIO/Image.t(127,9): error C2678: binary '<<': no operator found which takes a left-hand operand of type 'std::basic_ostream<char,std::char_traits<char>>' (or there is no acceptable conversion)Me:
(DIR) Post #AYpRvtFlZQ9VPG4OyO by mcc@mastodon.social
2023-08-17T03:29:36Z
0 likes, 0 repeats
Note: This may not make clear that the text above is only line one of what appears to be an eighty-line template error. And each of the eighty lines fills three lines of console scroll. I go back from Rust to C++ to compile ONE program and this is what I get
(DIR) Post #AYpRvts3H4mFJzyyeG by mcc@mastodon.social
2023-08-17T03:31:57Z
0 likes, 0 repeats
Wait what fucking cursed nonsense is this(Don't answer that, it's clearly some attempt to mix C macros with C++ template techniques that has gone horribly wrong)
(DIR) Post #AYpRvubmX3MBbpNUvY by mcc@mastodon.social
2023-08-17T03:37:10Z
0 likes, 0 repeats
Wait. Wait. What the devil.So at one level the problem here is that this code was written for a 2008-2012 version of GCC and I am now compiling it in a Visual Studio. The README warns it's never been tested on Windows.But that's not the real problem.The real problem is that these guys are <<ing an ostringstream into cerr.What fresh GNU-extension hell
(DIR) Post #AYpRvvXv2trSW8ZvXs by mcc@mastodon.social
2023-08-17T03:39:46Z
0 likes, 0 repeats
I guess it could be worse https://twitter.com/mcclure111/status/1104249530957090819
(DIR) Post #AYpRvwFAS6SKgGoSxM by mcc@mastodon.social
2023-08-17T16:42:37Z
0 likes, 0 repeats
Incidentally this is a real scary compiler warning. I don't like the implications.
(DIR) Post #AYpRvxDmoiwfiHAsRU by mcc@mastodon.social
2023-08-17T16:58:13Z
0 likes, 0 repeats
A totally generic observation: A thing that often frustrates me in compiler tools is when a thing says "can't X", and sometimes even explains why it can't X, but doesn't give an explanation of why it was *trying* to X. This is especially bad in convention-over-configuration or "magic" tools that try to guess your intention, which then sometimes guess wrong but don't make it clear how to convince it to guess a different way.It is also bad in CMake. But a lot of things are bad in CMake.
(DIR) Post #AYpRvyEB4kqupmMhgu by mcc@mastodon.social
2023-08-17T17:29:09Z
0 likes, 0 repeats
Project status: I successfully used CMake to build a C++ program! This took an embarrassing amount of time and I don't know whether to be embarrassed for myself or embarrassed for CMakeThe program generated this Cornell box. This… is not a very impressive Cornell box, honestly. I don't know what I expected.
(DIR) Post #AYpRvz3vyKFjQIa2Ma by mcc@mastodon.social
2023-08-17T17:33:47Z
0 likes, 0 repeats
To explain, what's happening here is I'm taking apart an integer-based ray tracing program implementation, with the intent of rewriting it to run on an FPGA. The FPGA doesn't have enough RAM to hold a framebuffer, so conventional tri rendering would be hard, but it *might* be fast enough to raytrace every single pixel just in time to shove that pixel out the door to the HDMI port just as the HDMI display expects it. Beam-chaser ray tracing. Raychasing
(DIR) Post #AYpRw02YKwk4SIwRqi by mcc@mastodon.social
2023-08-17T17:36:03Z
0 likes, 1 repeats
Why am I writing an integer raytracer for FPGAs, you may ask? Well, it's simple: I believe this is simpler than understanding the Vulkan hardware-raytracing-acceleration API and its inscrutable documentation. I'm… not sure that's a defensible statement. It didn't really occur to me to think about it this way until a few minutes ago. I think it might actually be true though.
(DIR) Post #AYpRw10oiswpTD8Zma by jbowen@mast.hpc.social
2023-08-17T17:37:25Z
0 likes, 0 repeats
@mcc > Why am I writing an integer raytracer for FPGAs, you may ask?That sounds like fun all on its own. No other reason needed.In a sonar lab I used to work at one of the guys wrote a raytracer in PostScript than ran (for weeks) on a printer. Because why the hell not?
(DIR) Post #AYpl1deWNuCDt86TvE by WAHa_06x36@mastodon.social
2023-08-17T19:17:29Z
0 likes, 0 repeats
@mcc I don't think I've ever felt that adding CMake to a project has ever made anything EASIER.
(DIR) Post #AYpl1eUdG9scUkU69A by mcc@mastodon.social
2023-08-17T19:18:50Z
0 likes, 0 repeats
@WAHa_06x36 the problem is that all conversations about replacing Cmake go like "why not just use autoconf?" "because I have to compile on Windows" "oh" and the conversation just ends there.
(DIR) Post #AYpl1fIyF0970s2Ibo by john@liberdon.com
2023-08-17T21:11:23Z
0 likes, 0 repeats
@mcc @WAHa_06x36 That's interesting. I haven't heard a tom of autotools boosterism. I usually hear Meson or (something incredibly niche like BJam) as "the" alternative. Some people really hand-write Ninja files, too. I've had medium experiences with SCons and tup, but I don't hear a lot of boosterism for either.
(DIR) Post #AYpltzSJXMpJ9AJJlh by mcc@mastodon.social
2023-08-17T21:21:11Z
0 likes, 0 repeats
@john @WAHa_06x36 Maybe Meson is really great. I don't know.I tried to use scons once because I like Python, and immediately ran into "this is not designed for any language except C/C++" and put it back down and didn't touch it again