slipher commented Oct 19, 2020

go ahead

illwieckz commented Oct 19, 2020

@slipher do you have something more to say?

slipher commented Oct 19, 2020

This is basically working as intended (see #202). You shouldn't be able to bring your own cgame when connecting to a remote server (unless cheats a…

slipher commented Oct 19, 2020

How about Warn with red text, for an extra big warning (and remove the redundant WARNING from the message).

necessarily-equal commented Oct 18, 2020

That makes me think… We could have force fields in games! Kinda like those in the Darth Maul duel in Star Wars 1 or the cargo bay access to space i…

Gireen commented Oct 18, 2020

I'm trying mapping and look what interesting things are possible. And thereby came up with a silly method to change maps. 😀

cu-kai commented Oct 17, 2020

Here is the gfxinfo output with the new GLEW...

illwieckz commented Oct 17, 2020

Oh, @cu-kai are you saying you get more than OpenGL 2 on the same macOS with that newer GLEW?

cu-kai commented Oct 17, 2020

Yes, this worked for me, thank you @slipher! It also means the game launches with an OpenGL 3.x context which I believe @illwieckz questioned on th…

slipher commented Oct 17, 2020

If you want to try it yourself, do brew install glew and replace the libGLEW.1.12.0 file (in the same directory as daemon) with /usr/local/Cellar/g…

slipher commented Oct 17, 2020

I reproduced the issue with a build based on the 0.52 branches. It occurs because GLEW provides null for the address of glMapBufferRange. I tried r…

slipher commented Oct 16, 2020

I tried with X instead of Wayland, but didn't managed to reproduce your bugs. I did find one other way for r_custom* to get changed. If you set r_m…

illwieckz commented Oct 16, 2020

Fixed twice by #385 and #377.

illwieckz commented Oct 16, 2020

Note: I use GNOME Shell with Xorg, not Wayland, maybe there is a difference between both?

illwieckz commented Oct 16, 2020

Hmm, I already knew GitHub does nasty things with commit ordering etc. but well, I missed two commits. I don't know yet for that resize issue, but …

slipher commented Oct 16, 2020

Are you sure you are testing with the right code? I pushed the r_allowResize-related commit later than the other ones.

illwieckz commented Oct 15, 2020

Also, because that's a development build, I'm running game as shared lib, I doubt this would have any impact, but that's to give more information a…

illwieckz commented Oct 15, 2020

How exactly do you produce bugs? Exactly the same way you attempted to: Alt+Enter, Alt+Tab and vid_restart… 😕 I'm using Ubuntu 20.04 with GNOME S…

slipher commented Oct 15, 2020

I did a lot of Alt+Enter, Alt+Tab and vid_restart on this branch in a Debian bullseye VM with GNOME desktop, but nothing bad happened. I also tried…

illwieckz commented Oct 14, 2020

rewrote comments

slipher commented Oct 14, 2020

I think imageDataSize is supposed to be the size on disk. For the unknown ones, I'd rather see a simple TODO: find imageDataSize rather than the my…

illwieckz commented Oct 14, 2020

fixed by #375

illwieckz commented Oct 14, 2020

SomaZ said the detection done in engine like OpenJK is about detecting file suffix (_nh) in a similar fashion to Darkplaces does, not about detecti…

slipher commented Oct 14, 2020

The meanings of 2 and 3 are swapped in the code vs. the above comment, but I suppose the order doesn't matter. So the code LGTM semantically. Pleas…

slipher commented Oct 14, 2020

LGTM; I too have no idea how this part of the code works. How did you run into this issue when there are no maps that use it?

illwieckz commented Oct 13, 2020

Hmm, since all those tricks are there to move the ParseMap call out of ASSERT, then we can just avoid a bool and do: if ( !ParseMap( &initialText,…

slipher commented Oct 13, 2020

There is already a facility for marking an unused variable, Q_UNUSED. It's used like Q_UNUSED(variable); Alternatively if(!parsed) ASSERT(false); w…

illwieckz commented Oct 13, 2020

@slipher, I rewrote the things accordingly. I added a UNUSED keyword to silent the compiler complaining about parsed being unused on non-debug code…

slipher commented Oct 13, 2020

Yes, Debian Bullseye

illwieckz commented Oct 12, 2020

@slipher On Mageia, RDX happened to contain a readable address, while on Debian Buster it contained 4. Did you mean Bullseye? If I'm right Buster…

illwieckz commented Oct 12, 2020

Thanks @mole99 for the testing! I updated a bit the script to properly catch any other crash that may happen and cleanup properly after that.

slipher commented Oct 12, 2020

I suggest as follows. Notes: DoDebugCode is used to execute the code only if the log level is debug. char buffer[1024] = ""; requires the entire b…

mole99 commented Oct 12, 2020

Can confirm that the patcher works for me.

illwieckz commented Oct 12, 2020

Here is an on-the-fly patcher (not destructing the original binary): #! /usr/bin/env bash # Unvanquished 0.51.1 Linux patcher # CC0 1.0 2020 Unvanq…

slipher commented Oct 12, 2020

Commands for fixing the client and server engine binaries from the 0.51.1 release: printf '\x8b\x74\x24\x20\x48\x31\xc0\x85\xf6\x0f\x94\xc0\xeb\x3f…

illwieckz commented Oct 12, 2020

Obsoleted by #341, @slipher did magic to make this not required.

illwieckz commented Oct 12, 2020

Fixed by #370.

illwieckz commented Oct 12, 2020

A daemon.log sample generated on Windows by @slipher shows that both GL_VENDOR and GL_VERSION strings are built-up the same way by the Nvidia propr…

illwieckz commented Oct 12, 2020

@slipher do you have any remark on it? I would appreciate some remark on pointer wizardry to keep an unmodified copy of **text if there is a better…

illwieckz commented Oct 12, 2020

This was tested on a large range of hardware and works as expected. Part of #381 can be rewritten in a better way once this is merged.

illwieckz commented Oct 12, 2020

Implementation is obvious.

illwieckz commented Oct 12, 2020

This fixes an obvious mistake and the implementation of this fix is obvious.

illwieckz commented Oct 12, 2020

@cmf028, do you have something to say about this?

briancullinan commented Oct 11, 2020

Nocrogiting: wasm works by way of LLVM, it would probably be more worthwhile to use that in the compile stack and leave wasm out of it. Same goes f…

illwieckz commented Oct 11, 2020

Anything making bots better is welcome! =) LGTM but I'm not a wizard on such topic.

Gireen commented Oct 11, 2020

I assume 0 is to not draw anything anytime? correct

Gireen commented Oct 11, 2020

func_spawn is an object/brush/mapentity that can be toggled. to appear or disappear by a trigger. I don't think that this is used in any of our map…

illwieckz commented Oct 11, 2020

I'm not sure to understand what it is about, do you have a simple in-game example?

illwieckz commented Oct 11, 2020

cg_drawBuildableHealth 1 = as before 2 = only for builders when damaged 3 = always for builders, for others when damaged 4 = for everyone when dam…

Gireen commented Oct 11, 2020

Now every combination is possible cg_drawBuildableHealth 1 = as before 2 = only for builders when damaged 3 = always for builders, for others when …

necessarily-equal commented Oct 11, 2020

just a style thing: this mixes tabs and spaces in the files it touches.

illwieckz commented Oct 11, 2020

It looks like to be an engine issue, most of the time is spent in Tess_SurfaceIQM from tr_surface.cpp: https://github.com/DaemonEngine/Daemon/blob/…

illwieckz commented Oct 10, 2020

Wow, that's really deep and interesting! I could make a binary patch for the 0.51 daemon to bypass the problem by replacing a few instructions in …

slipher commented Oct 10, 2020

I doubt there is anything to be done on the libstdc++ side. It is probably impossible to make a build that can work with both of the error_category …

illwieckz commented Oct 9, 2020

Wow, that's really interesting! Is there something to report on libstdc++ or we cannot do anything to make old builds like ours to be working again?

slipher commented Oct 9, 2020

The crash is caused by a mistake or shortcoming in the implementation of the "Dual ABI" concept in libstdc++. For this bug, the relevant dual-ABI c…

illwieckz commented Oct 8, 2020

Something like 5 or 8 minute for a default is good, I see no need for large time like 20 minute, 20 minute is already wrong to join even if no one …

Gireen commented Oct 8, 2020

instead of a timer that would be incremented everytime, we may just store a timestamp yes how long would the timeout then be? if its short like 8…

illwieckz commented Oct 8, 2020

instead of a timer that would be incremented everytime, we may just store a timestamp when both events occurs: game starts real player leaves a te…

illwieckz commented Oct 8, 2020

Another way to do it would be to reimplement g_emptyTeamsSkipMapTime in a way it does not use the game time but another timer that would reset and …

illwieckz commented Oct 7, 2020

Hmmm, I managed to load the game with a map on command line with a “fullscreen” game being minimized, and I cannot move the mouse. It's probably ju…

illwieckz commented Oct 7, 2020

So, to sum up the current status: input is lost when going from fullscreen to window (or back) [usually with Alt+Enter], → half fixed (display/hid…

slipher commented Oct 7, 2020

I fixed the thing that sets cvars when you toggle r_fullscreen. I'm happy enough with how it works now on my own machine. Later I'll try it in a Li…

illwieckz commented Oct 7, 2020

While running code from your PR, I noticed than when I cycle between windowed and fullscreen using [Alt+Enter], I can recover input by cycling betw…