Carrying the one true ring is a heavy burden (LOTR allusion intended).

Spirit has carried more luggage uphill than anyone (and frankly done a better job than any non-Spirit would have done).

Give him some slack. Yeah, in a perfect world many of the above posts shouldn't be in this thread. But we all know these kind of things happens.

I haven't had time to play Tronyn's apparently really good --- hey it has 200+ replies --- release.

Tronyn is in the top 3 of consistently awesome single player action and I won't spend 2 minute worrying about whether it is "Quake-enough".

I look for challenge, puzzles and entertainment value in release downloads. I also like a bit of surprise and some variation. Tronyn always delivers this. 

if you have a fast machine, they are pretty playable as is and deliver a solid 4~ hours of play. totally worth your time.

also, i feel i should apologize to tronyn for not noticing the maps were unvised. in hindsight, that i had to use r_flatlightstyles should have tipped me off.
and sorry to everyone who played and had their first playthrough of the maps spoiled by poor performance. 

used directq and didn't even notice the maps weren�t vis :\ 

Engine plays them fine as well. 

all those engines are pretty lightweight, but i can definitely see it chugging on graphically intense engines (dp being the obvious example, but there are a few more out there). 

There is always Quaketastic for hosting. Although that has a size limit that's possibly too low for some newer mods. 50 MB or so?

I bet Quakeone.com would also host some mods if one would ask them. I mean they host Reforged and they also kindly hosted the RMQ multiplayer demo.

Finally, there's always moddb, wordpress et al and I bet that Quakeone.com would kindly link to any such projects. 

Modern specs don't necessarily guarantee unvised excessive maps will run smoothly all the time. Different video cards may treat things differently it seems. It's what mh mentioned - the old Quake engine render wasn't meant for this and handles such stress inefficiently. 

How does the model mismatch bug happen? If I just use normal load/save and not quicksave, will I still encounter it? 

no idea. i never encountered it until after someone mentioned it and i went back to check.

someone said just saving and reloading again can fix it, dunno...
if the problem is indeed a random precache order in the mod, then i suppose reloading multiple times may get you the right precache order eventually? 

Will PM fix this bug then? 

After starting up QS fresh, most game loads from >= arcanum3 are broken here - just crazy mismatched models everywhere. This fixes it for me:

If game save is from arcanum5
"map arcanum4"
"changelevel arcanum5"

and then game loads work fine until i close the engine again. 

PM figured out what it was & it'll be fixed next release 

cool, thanks for the update!
if he doesn't mind sharing, i'd be interested in knowing what caused it, if only so i can avoid it myself? 

Due to the extra weapons, the mod precaches many models and sounds. I noticed that serverflags (i.e., runes) persist through the game, even across saves. The theory is I assign a custom weapon to a serverflag. If the serverflag is on, the weapon is enabled and precached. In other words, new weapons are available only when needed. Apparently, precaching controlled by serverflags does not work across saves. 

damn, that's clever too. shame it doesn't work though. at least in current times, saving on precaches isn't a problem anymore. 

For those who enjoy the extra detail of these things: this happens because serverflag IS saved in the save files, but it isn't applied until after precaches have been created.

Quake needs to rebuild the state of the server when loading a save. So the first thing it does is load the map and run the spawnframe and two physics frames. This creates all the static sounds, entities and makes sure that all the baseline initialisation is done.

It then has a complete snapshot of the world state in the save frame, entities, globals and so on, and proceeds to set the server up to match that exactly. The only problem is that models are saved by their modelindex, which is determined by the order that they are first precached in. This is a bit fragile, and can be broken in a few ways that I know of:

� Playing on a different version of a progs which precaches some extra models.
� Having a progs which randomises the precache order. Imagine an entity with a random selection of variant models to choose from. This can break the precache order if repeats occur in different patterns.
� PM's new variation, where a serverflag is set before the map is loaded when the save was made, but isn't reset by the loading code until after the precache table has been rebuilt.

Could anything be done about this particular case? It sounds tough. You'd have to encode any signal that serverflags were set in the save file, but you can't read anything from the save file until the precaches are built.

So best I think you could do is put some save game detection code in your mod. When you detect that
a) a save file has just been loaded
b) the serverflag was not set at worldspawn when you loaded this time.
and
c) the serverflag is set within the save file

then you know that the modeltable is broken, but that resetting the server will fix it.

The slightly ugly solution in this situation is then just to stuffcmd a restart in the console, so people who are loading under the broken savegames have to go through the loading process twice - once to set the serverflag to the correct value and a second time to get the modeltable correct.

You have to be REALLY careful here not to create an infinite loop of restarts by just throwing a restart in whenever you load a save and the serverflag is set. The way to do it is to store whether the serverflag is set in two variables, one of them flagged nosave and the other not. Set them both in worldspawn, and any time that you change the serverflag update them too. When you load a save the nosave variable will be set to the value the engine held BEFORE the save was processed and the other variable will be the value pulled from the save. A mismatch means you need to restart.


Preach thinks about it for a little bit...

Rats! It would all be fixed if you wanted to restart the server, but of course you actually want a save game to reload, and there isn't a server command for that. So the best you can do is run an objerror when you detect the corrupt table.
"Error: model table corrupt. Please reload your save to continue..."

This probably pushes it into territory awkward enough to not bother saving the feature. Interesting trip into the save game theory though... 

I already ripped out the serverflag precache nonsense from the mod. 

it'd be cool if an engine had an auto correcting system, storing like one of those error checking codes alongside the modelindex. if, on loading, it found the code didn't match, it would search through all the precaches for a matching one, and realign the modelindices.

...or just don't code stuff like that in the progs, i guess. hehe :P 

The problem is that this is a special case where we can identify how the precaches end up in a different order (value of a cvar) and we have a simple way to restore the conditions that created the original list (reset the cvar then load again).

In the case of random model selection, even if the engine could identify that the progs is doing that, there's still no way that it can recreate the random numbers generated in the original savegame. Saves don't record history like that, just an instant snapshot.

Of course the proper solution would be to recognise the importance of the modelindex table and store THAT in the savegame, but modifying the savegame format then creates incompatibilities between engines and savegames, which is worse. By the time you come up with an extensible "extended savegame file" which is saved side by side with the original save as "save0.svx" and probably stores all it's data in xml so that any engine can create new data there and the others can safely ignore it...you've probably gone too far. 

When's the re-release coming, Tronyn? 

Finally played this - though I can't give it a perfect score due to the slight inconsistency between the maps.

Actually, the pack could do better without the first map. The premise is good enough, it fits as an introduction but it doesn't deliver in execution and gameplay. Pulsar's map is great in its chosen textureset and it would remain perfect as a standalone creation (or in this context, as a secret map). The duo of map 3 and map 4 works wonders, they are without a doubt among the best I've played for the last two years - amazing interconnectivy, masterful brushwork, enjoyable setpieces. The finisher slightly deviates from the theme due to the tech bits but still feels right at home, and well, it looks brilliant. Too bad the flow remains sluggish till the boss fight.

Unlike Tronyn's previous maps which always turned to a chore at some point or another due to inhuman setpieces, I've managed to beat the pack on skill2 without too much frustration - definitely an improvement.

This will be even better with added maps and technical polish. Can't wait for the rerelease.

Keep up the good work guys. 

I just finished this pack. I gotta say the three last maps are part of a very small club of all time god like quake maps. The architecture especially on Dis Aliter Visum I is mind blowing.

Not to diss the other mappers, but I think this pack would have stood out more without the 2 'intro' maps. I agree with erc, Oderint Dun Metuat should have been released stand alone, or as part of another pack (for some reason I found the atmosphere very Quoth like, and a few Quoth monsters in there would have been fitting).

The first map had potential but seemed unfinished somehow, performance was atrocious without the excuse of Tronyns maps (incredibly complex brush work). Brushwork in the natural area (rocks and trees) looked like it was done in 10 minutes, and it really didn't need the ugly, framerate killing rain effect either. I also found the whole map pretty unbalanced monster/weapon wise. With quite a bit of polish I think it could have made a worthy intro map because the theme was well executed, very 'wizard like'. Maybe there could be medium sized 'blue' area at the end? 

I have read a few previous posts talking about a bug in the well.
Someone else mentioned "topedo-ing" out of there.
I found this impossible to achieve.
I held the forward key to make me dive all the way to the bottom, then hold it down to swim fast all the way to the top..nothing.
I have tried to do this along every inner surface of the well.
I have been trying to do this for the last 10-15mins, down then up etc.
This is the only thing that I am really upset about, as the rest of the game has been a challenge, but quite surmountable.

Apparently it IS possible to get out, but I just cannot find a way to jump out.
I'm not really looking forwards to having to start the map all over again, as it is huge, and I have already spent a lot of time on it.

Can anyone advise? 

bring down the console by pressing the � key (next the the 1 key),then type 'noclip' without quotes. Then press enter.

This will enable you to fly through walls, so you can just fly out of the place.

Repeat the process when you are back on dry land, to turn it off again. 

I just this minute figured out the noclip thing with some googling.
Thank GOD, imagine starting this map all over again from the beginning (aarrgghhh!!).