Just learned that there is no way for a GLSL shader to send data to the CPU side program.
Uncategorized
13
Posts
4
Posters
31
Views
-
@kertinker You should be able to write into SSBOs.
@landelare Oh, I hadn't learned about those. Thanks!
-
@kertinker You should be able to write into SSBOs.
@landelare Oh... sorry, I can't use storage buffer objects. GLEW_ARB_shader_storage_buffer_object is 0.
-
Just learned that there is no way for a GLSL shader to send data to the CPU side program. I think, at least. No way to emit debugging information at all, not even a puts.
At least if I understand, "Uniforms are so named because they do not change from one shader invocation to the next within a particular rendering call thus their value is uniform among all invocations."
Thought I could define a uniform, then set that to some debugging value in the shader.
@kertinker There is an extension for Vulkan that lets you use printf in GLSL shaders.
-
@landelare Oh... sorry, I can't use storage buffer objects. GLEW_ARB_shader_storage_buffer_object is 0.
-
@aeva @landelare Looks interesting! Hm...
printf("%d\n", GLEW_ARB_shader_image_load_store);
0
Looks like not that one either. GL_OES_mapbuffer maybe...? I might be hosed. -
@aeva @landelare Looks interesting! Hm...
printf("%d\n", GLEW_ARB_shader_image_load_store);
0
Looks like not that one either. GL_OES_mapbuffer maybe...? I might be hosed.@kertinker @aeva You don't printf into these. Use imageStore() if you want to use ARB_shader_image_load_store.
-
@kertinker @aeva You don't printf into these. Use imageStore() if you want to use ARB_shader_image_load_store.
@landelare @aeva I was using printf in a CPU side program, to print the "GLEW_ARB_shader_image_load_store" variable, which I think the OpenGL Extension Wrangler would set to 1, if ARB_shader_image_load_store was supported by my graphics card.
I made sure to call glewInit and have a valid context and all.
-
@landelare @aeva I was using printf in a CPU side program, to print the "GLEW_ARB_shader_image_load_store" variable, which I think the OpenGL Extension Wrangler would set to 1, if ARB_shader_image_load_store was supported by my graphics card.
I made sure to call glewInit and have a valid context and all.
@landelare as of version 1.1.0, GLEW signals extension availability with variables containing boolean values, so while the examples KT is posting look like name constants, they're mutable binary values that the initialization code sets up. relevant docs: https://glew.sourceforge.net/basic.html
@kertinker what GL version are you targeting and what platform? for a sanity check, could you post what GLEW_ARB_vertex_program is after init?
-
@landelare as of version 1.1.0, GLEW signals extension availability with variables containing boolean values, so while the examples KT is posting look like name constants, they're mutable binary values that the initialization code sets up. relevant docs: https://glew.sourceforge.net/basic.html
@kertinker what GL version are you targeting and what platform? for a sanity check, could you post what GLEW_ARB_vertex_program is after init?
@kertinker @landelare also if you happen to be on linux could you paste the output of glxinfo to a pastebin somewhere and link it here?
-
@landelare as of version 1.1.0, GLEW signals extension availability with variables containing boolean values, so while the examples KT is posting look like name constants, they're mutable binary values that the initialization code sets up. relevant docs: https://glew.sourceforge.net/basic.html
@kertinker what GL version are you targeting and what platform? for a sanity check, could you post what GLEW_ARB_vertex_program is after init?
This post is deleted! -
@kertinker @landelare also if you happen to be on linux could you paste the output of glxinfo to a pastebin somewhere and link it here?
@aeva @landelare Oh, sure. I don't have a "glxinfo" command, but visualinfo produces this: https://nopaste.net/l92XzrU5kr which... oddly includes GL_EXT_shader_image_load_store.
I'm clearly doing something wrong. Not sure how I could get a working OpenGL program running, without it saying extensions that are supported, are supported. Maybe you have to set a specific... mode, before they get supported?
Gli ultimi otto messaggi ricevuti dalla Federazione
-
@UkeleleEric @Lana I don't think "you" is exactly the right word here.
-
@luca yes, they did. Again...
-
@j_bertolotti LOL mice joke, where did you take this?
-
-
@LaVi temo che per indole politica Chuck Norris sarebbe stato più propenso ad aiutare Bossi che non a tirargli calci :(
-
@astro We always publish session notes on our website. If you are looking for the ones from our recent un-workshop, they are linked from the event page: https://fediforum.org/2026-03-growing-open-social-web/agenda/
-
To be honest, I haven't quite been feeling 100% since last night. My hands and feet are a bit cold, I've got a slight headache, and not much energy.
Whether it's just tiredness or a minor seasonal bug doesn’t really matter: an early night for me this evening, perhaps with a good film and a nice hot herbal tea.
-
User Repair of a Not User-Repairable Victron CCGX Issue
Power banks come in many sizes, and those that target construction sites are probably among the largest. The massive four ton unit based around lead-acid batteries which the [Buy it Fix it] YouTube channel got handed is a good example. Inside it are Victron CCGX inverters among a lot of other Victron electronics, with the control panel for the system throwing up an error that was deemed to be not user-serviceable. Naturally, this makes for a good challenge.
The exact error as thrown up on the central control panel is error #42, indicating a storage corruption issue on the device. According to the manual this means an issue with the internal flash memory that stores settings, serial numbers and WiFi credentials, requiring it to be shipped back to the manufacturer.
To further diagnose the issue, this Color Control unit was taken out of the power bank and coaxed onto a repair bench. This device has a whole host of Ethernet, CAN and other buses on the back, along with a USB host feature, but using the latter to reflash the firmware made no difference. Fortunately it’s just an embedded Linux system running on the System-on-Module and gaining remote SSH access was a snap due to easy root access.
Interestingly, running a diagnostic on the flash IC showed it to be still in good condition. Instead an ECC issue was logged that caused it to be marked as bad. This seems to have been due to the flash IC requiring 4 bits of ECC per 528 bytes, but the software using only a single bit. After reformatting and clearing the error it seems to have fixed the issue. Apparently it was just a weird configuration error that soft-bricked the device, raising the question of how that happened.
