Edge case maybe - but I record video at 720p fullscreen for my videos. My monitor is 1080 native of course. I use a hardware recorder over HDMI and cannot do 1080p at 60p. At 720 60p the videos look pretty good.

I'm glad I have the option of doing this. 

I play at fullscreen 720p in Quake these days. Just like the way it looks. 

On CRTs it makes a lot more sense to choose different resolutions. They can actually display them properly unlike LCDs and you might get better frame rate on your video card of the era. 

I currently play Quake on my laptop, which native resolution is 1200*800, but the closest match available in the video options is 1280*800. Would it be within the realm of possibilities to add a 1200*800 mode in a future update? 

Why on earth would you want to switch to a lower res fullscreen mode

The absolute maximum res I'd play quake on is 720p.

Old games like quake often look better when played at lower resolutions.

The original id levels for example, really don't improve with resolution: after a point, all the higher-res does is look increasingly incongruous with the level art. At 720p, you are already way past that "wall" for old-skool quake maps.

720p is pretty good for AD-style modern maps.

/opinion. 

No matter what I do, I can't seem to get non-blurry textures in the latest QS (and QSS, for that matter). I do have gl_texturemode set to GL_NEAREST_MIPMAP_LINEAR in my config and I even duplicated the line in an autoexec to be safe, but to no avail. Thoughts? 

gl_texture_anisotropy 1 I think? (can't check easily, on mobile) 

it worked. My config was actually saved from a previous install, I suppose one that I used with hi-rez textures and linear_mipmap_linear. Anisotropy was set to 16. Thanks Eric! 

When you play a demo and press tab to show the stats, the skill displayed is not the one the demo was recorded with but the one selected (or not) by the watcher. Is this normal behavior? It would make a lot more sense to show the actual skill set for the recording. 

Mark V is the only engine that currently can do that it (and has done it for 5 years?). When the skill level changes, it broadcasts a hint. This is stored in the demo because it is part of the network code, so it is displayed during demo playback and cooperative players status bar is updated in real time.

And it is done in a way that non-supporting clients just ignore. (A Spike trick).

Otherwise, it is impossible. A demo does not normally know what skill level is selected. That information is ordinarily lost. 

good bug though, the scorebar should probably display no skill at all during demo playback, if it is not known. 

one thing i noticed on older engines is that stuffcmd gets sent to clients when playing demos. may have been fixed in newer engines. i supressed it in mine. 

one thing i noticed on older engines is that stuffcmd gets sent to clients when playing demos. may have been fixed in newer engines. i supressed it in mine.

"bf" is a stuffcmd.

You sure want stuffcmds in the demos ;-)

That's just one example, but removing stuffcmd from demos isn't a fix. Yet at the same time, stuffcmd can do toxic things in demos too.

I get what you were wanting to address -- especially a demo where server is sending key binds, but removing stuffcmd from demos isn't how to resolve it. FTE has things sandboxed some. Mark V less than FTE, but views keybinds not done by the user as temporary until disconnect (and that includes demos), and stored in different field that never saves to config. 

Though I must admit i don't typically do this for quake, I play a lot of older games 'pillarboxed' on my widescreen monitor with a 4:3 resolution. Letting users choose things like this is important for edge cases likr that.

Maybe i should try some 4:3 Quake, actually... 

"Fixed: Alias commands are not stuffed to the client when watching demos."

i dug back through my whatsnew.txt It's where i was watching some old demos and noticed i had these weird alias commands; cant remember what demo (runequake?) but kinda annoying. Stuffcmd isnt specifically blocked.
``
void Cmd_Alias_f (void)
{
cmdalias_t *a;
char cmd[1024];
int i, c;
char *s,*n;

if (cls.demoplayback)//R00k dont stuff alias commands when watching a demo.
return;
``
trivial. 

aye, aliases in demos is annoying.
as is the engine silently ignoring the alias commands that you're typing simply because there's a demoloop playing in the background...
ignoring aliases completely means that mods that use them to slightly reduce network traffic cause error spam.

urgh, now I have to resist the urge to make a 'trap' demo that uses aliases to prevent any attempt to stop playing that demo. 

Hmmm, I see your point; 'silently' confuses the end user. For example, loading up the game, demo is playing, type exec frikbot.cfg..
"Where's my frikbot aliases?!"... 

Wanted to make a coop server for ad_sepulcher, got into the same issue as ericw up there: lots of "packet overflow", some sounds weren't playing, etc..
Are there plans on making the unreliable message limit something changeable on the server configuration or something? Elsewise, what could I use to host a coop server reliably? I've never hosted one before, what do you guys use? 

Netquake is just not reliable enough for mods like AD. You should turn particles and shell ejection off (read the AD readme for how to do this). You should try playing coop with QuakeSpasm-Spiked, which has improved networking.

Ideally the Quake community should have rallied around the QuakeWorld protocol when the source was released, and backported compatability with the netquake progs. Alas we have a shedload of NQ engines which are pretty useless at multiplayer. 

Yeah, if you want to play AD coop then you really ought to upgrade from QS to QSS.
FTE has all the same networking improvements but I accept that its also much easier to configure wrongly...
DP also has many of the same improvements, but its insistence on float coords combined with its inability to split baselines means that its unable to run ad_sepulcher. Configuring it to use the bjp3 protocol would theoretically work if its support for bjp3 was not buggy.

The vanilla QuakeWorld protocol was quite poo and quite incompatible with NQ, yes it would have been nice to have a single protocol that everything else was compatible with, but really that's just wishful thinking. It would have broken more than it would have solved.
Do note that the FTE stuff I ported over to QSS is not like vanilla quakeworld at all - its more like dpp7. It won't cope with packetloss quite so well, but it will cope MUCH better with massive entity counts, and without requiring serverside translation of QC's writebytes (this is one of the more annoying things I had to add to FTE to get it to run NQ mods properly, and I'm very glad other engines don't need anything equivalent, and no way was I going to add that mess to QSS too). 

@spike @c0burn:
Thanks for the pointers. I've tried using QSS for the server but the following happens: https://pastebin.com/kc54XVpM. Apparently even with the dedicated flag it still tries to capture mouse? The SDL warning message is repeated constantly, every line omitted is that exact message. Trying to enter ad_sepulcher crashes the clients (both QS and QSS) at the tune of https://pastebin.com/dfK1c4s9, both of the clients outputting this, and they get stuck at the "loading" screen, so I have to kill them. dpkg -l *libsdl* outputs this: https://pastebin.com/4vnHjrNz, so I gather I have installed SDL2 correctly (?). Other maps also output warnings about misaligned lumps but they are playable.
I've found this problem on Ubuntu, if I'm not misremembering 16.04 (not "my" PC), and on a Debian 9, both fairly vanilla installations. The misaligned lumps problem seems separate from the SDL warning (apparently it's just the program trying to grab mouse and keyboard input, but since it's running in -dedicated mode it's pointless), and the former is what I gather is causing these problems, rendering the map unplayable.
I've also tried running the server from QSS (not dedicated, in-game), but a friend couldn't connect either; same misaligned lumps message was seen on his end. This was all using the version downloadable from http://fte.triptohell.info/moodles/qss/, didn't try compiling the latest in the repo nor patching QS with the change @ericw posted. Will probably try tomorrow, if I have the time (but I gather that @ericw's change is undesirable in some way, if I didn't misinterpret @spike's comment, but I can't say I internalized quake's protocol). 

ignore the misaligned lumps. that doesn't affect x86 at all (but is fatal for other cpus).

Sorry about the grabs spam. It seems the linux version still uses SDL1 for some reason, while the win32+SDL2 build doesn't show that there's anything stupid happening. Should just be a problem due to the spam, but yuck, sorry.

I've no idea why its locking up for you.
Obviously its not meant to do that... 

In the end we ended up playing with ericw's patch for vanilla QS. No issues there.
All in all, that patch was a gateway drug! :P The code is easier to read than I first thought it'd be, so now I'm adding 6DOF. If I'm not mistaken this is something that can't be implemented at a qc level, right? I have but a passing familiarity with quakec, so maybe this could be less client-dependent and I'm not realizing it.
Either way, having much fun with this, keep up the good work. 

Any hope for incorporating some of the additions to the Arcane version of Quakespasm/QSS? Like the mod menu. 

So is pampa the anon mystery guy that wrote the 6DOF version of FitzQuake several years ago and then Spike added the modification to FTE?